US 8414921

I'll run targeted searches for this specific patent number and its litigation history.

Now let me search specifically for CAFC 2026 docket activity and the litigation history.

Let me look specifically for the Federal Circuit appeal docket and the West Virginia decision on the '921 patent.

US Patent 8,414,921 — Analyst Summary

Verification approach: I searched for the literal string "8414921" and its formatted forms. Results were consistent with the authoritative full text supplied in my working record. I did not find any search result that assigned a different patent to this number, and no result for a "similar" number was substituted. Where I could not confirm a source directly, I say so below.

---

1. Bibliographic data

Note: the Google Patents record shows the priority date as 2005-12-16 and the filing date as 2006-12-12, and lists the US "priority" of the national-stage application itself as 2006-12-12 — this reflects national-stage entry mechanics, not a separate earlier US filing.

---

2. Abstract (verbatim)

"Disclosed are pharmaceutical compositions comprising fixed-dose combinations of a dipeptidyl peptidase-4 inhibitor and metformin, methods of preparing such pharmaceutical compositions, and methods of treating Type 2 diabetes with such pharmaceutical compositions."

---

3. Plain-language overview of the independent claims

There are five independent claims: 1, 10, 16, 20, and 21. Claim 1 is the composition claim that has driven the litigation; claims 22–28 are product-by-formula claims tied item-by-item to the worked Examples.

Claim 1 — the core composition claim (weight-percent envelope).
A pharmaceutical composition containing, by weight: (a) ~3–20% sitagliptin (or a pharmaceutically acceptable salt); (b) ~25–94% metformin hydrochloride; (c) ~0.1–10% lubricant; (d) 0–35% binder; (e) 0.5–1% surfactant; and (f) 5–15% diluent. The two narrow numeric windows in (e) and (f) — a very small surfactant range and a mid-size diluent range — are what distinguish it from the generic DPP-4/metformin combination concept.

Claim 10 — unit-dose composition claim.
A composition defined by absolute doses rather than percentages: sitagliptin at 25–200 mg per unit; metformin HCl at one of the FDA-approved strengths (250, 500, 625, 750, 850, or 1000 mg); plus ~1–2% lubricant, ~7% binder, ~10% diluent, ~0.5% surfactant.

Claim 16 — first method claim.
A method of treating Type 2 diabetes in a human by orally administering the composition of claim 1.

Claim 20 — second method claim.
A method of treating Type 2 diabetes in a human by orally administering the composition of claim 12 (i.e., the unit-dose claim limited to sitagliptin 25/50/75/100/150/200 mg with metformin 500/850/1000 mg).

Claim 21 — "consisting essentially of" counterpart to claim 1.
Same ingredient list and same six numeric ranges as claim 1, but using transitional phrase "consisting essentially of," which narrows the claim by excluding additional unrecited active ingredients that would materially affect the basic characteristics of the combination.

Representative dependent claims: claim 3 (narrowed wet-granulation-style envelope: 5–18% sitagliptin, 65–77% metformin HCl, 1–2% lubricant, 4–9% binder); claim 4 (magnesium stearate or sodium stearyl fumarate + PVP); claims 5–8 (specific 50/500 and 50/1000 potency formulations); claim 9 (dihydrogenphosphate salt); claim 11 (SST/PVP/MCC/SLS identity of excipients); claims 12–14 (50 mg sitagliptin with 500, 850, or 1000 mg metformin); claim 15 (tablet form); claim 18 (prepared by wet granulation); claim 17 (flavoring/colorant/sweetener). Claims 22–28 recite complete tablet formulas matching Examples 1–7.

---

4. Litigation and 2026 docket activity

District court. Google Patents' litigation feed links five US district court cases to this family:

• D. Del. 1:19-cv-00315, 1:19-cv-00316, 1:19-cv-00317, 1:19-cv-00319, 1:19-cv-00347 (all 2019)
• D. Del. 1:20-cv-01099 (2020)
• N.D. W. Va. 1:19-cv-00101, Merck Sharp & Dohme LLC v. Mylan Pharmaceuticals Inc. et al. In that case the court (Judge Irene Keeley) decided infringement of claim 1 of the '921 patent (alongside claim 3 of the '708 patent). The court's opinion (Dkt. 204, filed 26 Oct 2022) records that the '921 patent relates to the sitagliptin/metformin co-formulation and notes Mylan did not contest infringement of claim 1 of the '921 patent — Merck prevailed.

Federal Circuit. News coverage dated 18 September 2026 (pharmaphorum, "Merck defends blockbuster Januvia franchise from patent challenge") reports:

• The Federal Circuit affirmed the USPTO's decision upholding US 7,326,708 ('708, the sitagliptin dihydrogen-phosphate salt patent), on Viatris/Mylan's appeal from the PTAB.
• Separately, "last week" the district court in West Virginia ruled for Merck on a case that "challenged the '708 patent … as well as one other (No. 8,414,921)."

Important caveat — do not over-read this. I could not confirm from any authoritative source that a 2026 Federal Circuit docket exists in which the '921 patent itself is the patent on appeal. The reported CAFC activity concerns the '708 patent IPR appeal. The '921 patent's 2026 win appears to be at the district court level in N.D. W. Va. (per pharmaphorum), and the same article states Viatris "could still take its cases further" — i.e., an appeal of the '921 ruling may or may not have been filed, and I found no docket number for one. I also note the Owler/Endpoints-style account of the West Virginia and Federal Circuit wins uses Q2-2021 comparative revenue figures, which suggests that text was originally published around 2021–2022 and may have been re-surfaced/re-dated; the 2021 vs. 2026 timing of these two rulings is therefore not something I can resolve with confidence.

Other proceedings of record for this family: EPO oppositions (Apotex, 2011; Teva, 2008 — maintained as amended); CJEU reference on Janumet SPC validity (Finnish court referral, AG opinion 6 Jun 2024); multiple settlements with generics allowing US entry May/July 2026. A US PTAB petition document (ptacts.uspto.gov petition 1537262) quotes Merck's complaint language asserting that Watson's ANDA product infringes "at least claim 1 of the '921 patent."

---

5. Data-quality notes (stated literally, not corrected)

Per the operating rule to interpret identifiers literally, I flag the following defects present in the patent text as published:

• Claim 25 recites "sitagliptin free base anhydrite"; claims 27 and 28 recite "free base anhydrite." The correct term elsewhere in the patent is "anhydrate."
• Example 7 states sitagliptin phosphate monohydrate 128.5 mg is "*Equivalent to 50 mg of sitagliptin free base anhydrate" — inconsistent with the specification's own conversion table (128.5 mg ↔ 100 mg), and with the Example 6 entry, which uses the identical 128.5 mg figure for 100 mg free base.
• The [*] markers in claims 22–28 are footnote markers carried over from the example tables; the underlying text in the claim body is unaffected.
• The Google Patents page shows a "US 8572206 A" cross-reference in the metadata block, which does not correspond to the application number 12/085,722 given for this patent; I am not able to explain that field and would not treat it as authoritative.

---

6. Confidence statement

• Bibliographic data, claims, and abstract: high confidence — taken from the authoritative full patent text supplied to me, corroborated by PubChem patent record US-8414921-B2 and DrugPatentWatch.
• Orange Book / Janumet linkage and 2028/2029 expiries: high confidence — multiple independent aggregators agree.
• District court litigation list and the West Virginia outcome: moderate-to-high confidence — docket IDs are from Google Patents' litigation feed; the West Virginia opinion text is from a court document PDF.
• Existence of a CAFC 2026 docket for the '921 patent specifically: not confirmed. What I can support is a 2026 Federal Circuit affirmance concerning the '708 patent, plus a 2026 West Virginia district court ruling covering both '708 and '921. I would verify on PACER/CM-ECF and the Federal Circuit docket before representing that the '921 patent is on appeal.

I'll research litigation involving US patent 8414921. Let me run several searches.

The first result set confirms the Orange Book listing. Now let me search specifically for the litigation cases linked on the patent page.

Good progress. Now let me pin down which specific cases asserted the '921 patent and their outcomes.

Litigation Involving U.S. Patent No. 8,414,921

Patent context: US 8,414,921 ("the '921 patent"), Pharmaceutical compositions of combinations of dipeptidyl peptidase-4 inhibitors with metformin, is Merck's Janumet® formulation patent (sitagliptin phosphate + metformin HCl tablets). It was listed in the FDA Orange Book against NDA 022044 (Janumet), with a drug-product expiry of July 21, 2028 and pediatric exclusivity to January 21, 2029 (per the FDA Orange Book entry and the Google Patents page at https://patents.google.com/patent/US8414921/en).

The '921 patent has been litigated — almost entirely as part of the sitagliptin phosphate ANDA wave of 2019 and the resulting multidistrict litigation. Note an important limitation on the scope: according to the JPML transfer motion, Merck asserted the companion '708 patent (sitagliptin dihydrogenphosphate salt) against 23 defendants in 13 groups, but asserted the '921 patent against only 3 of those defendant groups, including Mylan. So not every case below necessarily put the '921 patent at issue.

---

1. The core proceeding: MDL No. 2902

In re: Sitagliptin Phosphate ('708 & '921) Patent Litigation, J.P.M.L. MDL No. 2902.

• Plaintiff/patentholder: Merck Sharp & Dohme Corp. (now Merck Sharp & Dohme LLC)
• Defendants: the 14 generic ANDA filers listed in Schedule A of the transfer order
• Jurisdiction: consolidated pretrial proceedings transferred to the District of Delaware (Wilmington), assigned to Judge Richard G. Andrews
• Transfer order: August 8, 2019
• Status: MDL closed/consolidated; the constituent cases proceeded individually (many settled; the Mylan action was tried in West Virginia — see below)
• Source: JPML transfer order, https://www.jpml.uscourts.gov/sites/jpml/files/MDL-2902-Initial_Transfer-07-19.pdf ; and https://www.courtlistener.com/docket/17279840/56/in-re-sitagliptin-phosphate-708-921-patent-litigation/

A key pretrial ruling construed the '921 patent: Merck Sharp & Dohme Corp. v. Anchen Pharmaceuticals, Inc., No. 1:19-cv-00311-RGA (D. Del.), Markman opinion filed November 17, 2020 (D.I. 135), construing claim 1 of the '921 patent (percentages of sitagliptin, metformin HCl, lubricant, binder, surfactant, and diluent) and reserving judgment on "surfactant," "sitagliptin," and "sodium lauryl sulfate." Source: https://cases.justia.com/federal/district-courts/delaware/dedce/1:2019cv00311/[67784/135](https://assignmentcenter.uspto.gov/search/patent/reelFrameDetail?reelFrame=67784-0135)/0.pdf

2. The underlying district court complaints (D. Del., Judge Andrews)

Filed February 13–19, 2019; all later folded into MDL 2902:

Of these, the complaints I could verify as expressly pleading both the '708 and the '921 patents include Sun Pharma (C.A. 19-319, referencing ANDAs 202423, 205078, 207823 for Januvia/Janumet XR/Janumet) and Teva (19-318). Source: Sun complaint, https://www.docketalarm.com/cases/[PTAB](/ptab)/IPR2020-01072/Sun_Pharmaceutical_Industries_Ltd._v._Merck_Sharp_%26_Dohme_Corp/docs/07-10-2020-Patent_Owner/Exhibit-2040-EX2040___Sun_Complaint.pdf

3. The case that went to trial: Merck v. Mylan (N.D. W. Va.)

• Case: Merck Sharp & Dohme Corp. (later LLC) v. Mylan Pharmaceuticals Inc. (and Mylan Inc.)
• Jurisdiction: U.S. District Court for the Northern District of West Virginia, Judge Irene M. Keeley
• Case No.: 1:19-cv-00101-IMK
• Filed: May 2, 2019 (after Merck voluntarily dismissed its parallel Delaware action, 1:19-cv-00315)
• Patents asserted: the '708 patent (claim 3) and the '921 patent (claim 1)
• Procedure: five-day bench trial; post-trial briefing
• Outcome: Merck won. In a decision announced September 22, 2022 (written opinion entered October 26, 2022, D.I. 204), the court held both patents valid and infringed by Mylan's ANDA products.
• Status: appealed by Mylan/Viatris to the U.S. Court of Appeals for the Federal Circuit. (I was unable to confirm the appellate case number or final disposition from the sources retrieved, so I will not state an appellate outcome.)
• Sources: Merck press release, https://www.merck.com/news/u-s-district-court-rules-in-favor-of-merck-in-sitagliptin-phosphate-patent-lawsuit/ ; opinion at http://ocr.docketalarm.com/cases/West_Virginia_Northern_District_Court/1--19-cv-00101/Merck_Sharp_%26_Dohme_LLC_v._Mylan_Pharmaceuticals_Inc._et_al/docs/204.pdf

4. Additional cases flagged on the patent page but not fully verified

The Google Patents page for US 8,414,921 also lists, as related litigation:

• D. Del. 1:19-cv-01489 — details not confirmed in my searches
• D. Del. 1:20-cv-01099 — details not confirmed in my searches

(These are consistent with later-filed ANDA actions against additional generics; I could not verify the parties or that the '921 patent specifically was asserted.)

5. Related but distinct proceedings (for context — not '921 cases)

To avoid conflating matters: several high-profile sitagliptin proceedings concerned only the '708 salt patent, not the '921 patent:

• IPR2020-01072 (Sun Pharmaceutical Industries Ltd. v. Merck) — challenge to the '708 patent; Merck prevailed at the PTAB (May 7, 2021), and Viatris appealed to the Federal Circuit.
• Merck v. Dr. Reddy's Laboratories, D. Del. 1:20-cv-00847 and 1:21-cv-01033 — settled (2022), injunction against infringement.

6. Overall current status (as of April 26, 2026)

• The '921 patent is in force and Orange Book–listed, expiring January 21, 2029 including pediatric exclusivity.
• The 2019 wave of '921/'708 ANDA cases was resolved largely by consent judgments and settlement agreements under which generics may launch Januvia/Janumet in May 2026 (or earlier under certain circumstances) and Janumet XR in July 2026.
• The Mylan/Viatris case is the one substantive merits decision on the '921 patent (valid and infringed), and it was appealed; I could not verify the appellate disposition from the retrieved sources.

---

Confidence / caveats: The existence, courts, case numbers, and parties for the 2019 D. Del. cases and the MDL are well-grounded in the JPML order and docket materials above. The precise mapping of which of those cases asserted the '921 patent (vs. only the '708 patent) is confirmed only for Mylan, Sun, and Teva; the JPML record states '921 was asserted against just three defendant groups. The Federal Circuit outcome of the Mylan appeal and the subject matter of 1:19-cv-01489 and 1:20-cv-01099 are not confirmed and should be verified against PACER/CourtListener before being relied upon.

I'll verify the ODP default against public sources before writing this up.

Key signal: the sitagliptin IPR I've found so far (IPR2016-01131/01132, and the Viatris appeal) targets the '708 salt patent, not the '921. Let me confirm whether any petition names the '921 specifically.

Critical lead: IPR2020-00040 (Mylan v. Merck). I need to pin down which patent it targeted and its outcome.

PTAB / AIA Trial Proceedings — US 8,414,921 B2

Prepared for: a defendant evaluating assertion of U.S. Pat. No. 8,414,921 ("the '921 patent")
Patent: Pharmaceutical compositions of combinations of dipeptidyl peptidase-4 inhibitors with metformin — Merck Sharp & Dohme LLC, app. 12/085,722, priority 2005-12-16, granted 2013-04-09
Data cutoff: 2026-09-28

---

Proceedings overview

There are no AIA trial proceedings on US 8,414,921. The structured "PTAB proceedings on file" block from the USPTO Open Data Portal returns zero IPR, PGR, or CBM cases, and my independent web checks surfaced no petition against the '921 patent — instituted, denied, or settled. The one AIA trial in this patent family, IPR2020-00040, targets the different Merck patent US 7,326,708 (the '708 salt patent), with three joinder IPRs (Teva, Dr. Reddy's, Sun) all riding on that same '708 challenge.

The bottom line for a defendant: the '921 patent has never been to the Board. It has not been tested, narrowed, or canceled in any AIA forum. It has, however, been litigated to a merits judgment in district court — claim 1 was held valid and infringed after a five-day bench trial (N.D. W. Va. 2022), and the Federal Circuit affirmed the sibling '708 patent's PTAB survival in 2022. So this is not a "survived two IPRs and is hardened" story and it is not a "claims are canceled, the troll has no case" story. It is a never-challenged-at-the-PTAB patent backed by a district-court validity judgment and a patent family that has already beaten an IPR once. Your § 102/§ 103 runway on the '921 is statutorily clean; the practical risk is discretionary denial, § 315(b) timing, and issue preclusion rather than estoppel.

Breakdown by status: active — 0 · claims invalidated — 0 · claims sustained — 0 (no FWD on this patent) · settled — 0 · institution denied — 0.

---

Related AIA proceedings in the same family — NOT on the '921 patent

I include these because they are the proceedings a defendant will be told about (and will find when searching "sitagliptin IPR"), and because conflating them with the '921 is the single most likely analytical error in a defense memo. None of them challenged a claim of the '921 patent.

IPR2020-00040 — Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp.

• Type: Inter Partes Review
• Filed: 2019-10-30 (petition date, per the Final Written Decision)
• Patent challenged: US 7,326,708 B2 — not the '921 patent
• Status: Final Written Decision — "Determining No Challenged Claims Unpatentable"; closed 2021-05-07. (Petitioner's framing: patentable.)
• Judge panel: Sheridan K. Snedden, Robert A. Pollock, Timothy G. Majors (APJs)
• Petition grounds: Claims 1–4, 17, 19, and 21–23 of the '708 patent were challenged on (i) § 102 anticipation by WO 03/004498 / U.S. 6,699,871 (collectively "Edmondson"), including an inherent-anticipation theory that reacting sitagliptin with phosphoric acid yields the 1:1 dihydrogenphosphate ("DHP") salt "every time"; and (ii) § 103 obviousness over Edmondson in view of Bastin (salt selection) and Brittain (hydrates/solvates).
• Institution decision: instituted 2020-05-12, on all asserted grounds. The panel instituted on the inherency theory despite finding no express disclosure of the 1:1 salt, based on the preliminary record suggesting the 1:1 salt was a "necessary byproduct" of contacting phosphoric acid with sitagliptin.
• Final Written Decision (2021-05-07): no challenged claim held unpatentable.
  • Anticipation: The panel found Edmondson discloses only two "closed lists" — 33 compounds (including sitagliptin) and eight "[p]articularly preferred" acids (including phosphoric acid) — and held Mylan could not bridge the gap via In re Petering's "at once envisage" doctrine. On appeal the Federal Circuit credited Merck's count that these lists yield 957 putative salts, some of which "may not even form under experimental conditions," "a far cry from the 20 compounds 'envisaged' by the narrow genus in Petering."
  • Inherency: rejected. "Evidence, both experimental and from the technical literature, undeniably showed that 1:1 sitagliptin DHP does not form every time stagliptin and DHP were reacted." The panel also criticized Mylan's expert (Dr. Chyall) for making "at least eight changes" to the WO '498 example — swapping phosphoric acid for the disclosed hydrochloric acid — so the experiment was "much different than the 'mere swapping' of acids."
  • Obviousness: rejected. Merck successfully antedated Edmondson for claims 1, 2, 17, 19, 21–23 (prior reduction to practice), converting Edmondson from a § 102(a) reference to a § 102(e) reference and then excluding it under pre-AIA § 103(c)(1) common-ownership. For claims 3 and 4 (no antedating argument), the panel found no motivation with a reasonable expectation of success.
• Settlement / termination: none. The case was decided on the merits.
• Joinders (all to this proceeding, all on the '708 patent):
  • IPR2020-01045 (Teva) — the "Viatris/Teva" challenge referenced in Merck's public statements
  • IPR2020-01060 (Dr. Reddy's Laboratories, Inc. and Dr. Reddy's Laboratories, Ltd.)
  • IPR2020-01072 (Sun Pharmaceutical Industries Ltd.)
• Appeal: Mylan Pharms., Inc. v. Merck Sharp & Dohme Corp., No. 2021-2121 (Fed. Cir. Sept. 29, 2022) — affirmed on substantial evidence as to both anticipation and obviousness. Mylan noticed appeal 2021-07-08. Merck's 2022-09-22 press release separately states that Viatris had appealed the Board's May 7, 2021 decision; I could not verify a separate Viatris/Teva docket number, and I will not guess one.
• Defensive value: None against the '921 patent. Zero. Not one '921 claim was challenged, so no § 315(e)(2) estoppel attaches to the '921, and the FWD's cancellation certificate cannot have touched it. Its real significance is negative: it is a proof point that Merck has already defended a sitagliptin-family formulation patent to a PTAB win and a Federal Circuit affirmance, and it maps out Merck's successful antedating/common-ownership playbook. Expect Merck to run the same script.

---

Proceedings that are not PTAB proceedings (do not misread the docket flags)

The litigation links in the source data — D. Del. 1:19-cv-00315, -00316, -00317, -00319, -00347, -01489; D. Del. 1:20-cv-01099; N.D. W. Va. 1:19-cv-00101 — are Hatch-Waxman ANDA infringement suits consolidated as MDL No. 19-2902-RGA, In re Sitagliptin Phosphate ('708 & '921) Patent Litigation (D. Del., Judge Richard G. Andrews). They are district court matters, not AIA trials. Likewise, the Unified Patents portal links are just that portal's aggregation of district court dockets — there is no defensive-aggregator IPR in this chain (no Unified Patents, RPX, or similar petitioner).

Two district court events genuinely matter to a '921 defendant:

1. Markman (2020-11-17) — the MDL court construed '921 terms including "sitagliptin" and "surfactant" (claims 1, 3, 5–8, 10, 11, 21 of the '921 were identified for construction). Notably, "due to the pending IPR in connection with the '708 patent, Mylan did not join in any of Defendants' proposed claim constructions."
2. Bench trial judgment (N.D. W. Va. 1:19-cv-00101-IMK, Judge Irene Keeley) — Merck's 2022-09-22 announcement: the court found both the '708 and the '921 patents "valid and infringed"; the sealed Memorandum Opinion and Order Following Bench Trial is docketed at Dkt. 204 (filed 2022-10-26, public version on CourtListener). Only claim 1 of the '921 patent was at issue.

---

Strategic summary

Claim status. Per the ODP structured record, all 28 claims of US 8,414,921 remain in force — none canceled, none amended, none confirmed-by-certificate, because no AIA trial has ever been instituted. Claim 1 (the composition claim with "about 3 to 20% by weight of sitagliptin … about 0.5 to 1% by weight of a surfactant") and the claim-22–28 tablet claims tied to the exact Example 1–7 formulations are the commercially operative ones for JANUMET®. Claim 1 is UNTESTED at the PTAB — but it is judicially tested: it was held valid and infringed after a full bench trial. That is a materially different posture from "untested." Any invalidity theory you intend to run must be able to explain why it succeeds where Mylan's did not.

Estoppel landscape. § 315(e)(2) estoppel is not a constraint on you with respect to the '921 patent. Mylan, Teva/Watson, Dr. Reddy's, and Sun are estopped only as to claims of the '708 patent on grounds raised or reasonably raisable in IPR2020-00040 and its joinders. Grounds against the '921 patent's composition claims — including formulation-art combinations (e.g., the prosecution-rejection references US 2005/0051922, US 2003/0166578, US 2009/0253752, and '810) and the surfactant-as-wetting-agent rationale — remain fully available to any petitioner, including the '708 IPR petitioners as to the '921. Conversely, the fact that the '708 IPR petitioners did not petition on the '921 means you carry none of their baggage, and they carry none of yours.

Pattern signals. (1) No repeat-petitioner pattern on this patent — one family IPR, three joinders, all aimed at the compound/salt patent, not the formulation patent. (2) Merck defends aggressively at the PTAB and has now won there twice over (FWD 2021-05-07; Fed. Cir. affirmance 2022-09-29), and it litigated the '921 to a merits win rather than settling. (3) No defensive aggregator in the chain. (4) The clinical/legal clock has largely run: the '921 has an adjusted expiration of 2028-07-21 (with pediatric exclusivity taking Orange Book protection to 2029-07-21 per Merck's own statement), and Merck has already signed settlements permitting generic JANUVIA/JANUMET entry in May 2026 (earlier in certain circumstances). As of today (2026-09-28) that date has passed — which sharply reduces both the ROI of an IPR and your leverage in negotiating one.

---

Recommended next steps

1. Do not file an IPR on the '921 patent reflexively. There is no PTAB precedent to piggyback on, the patent survives to 2028-07-21 (pediatric to 2029-07-21), and Merck has a 2022 district-court validity judgment on claim 1 that any panel will read against you. Cost/benefit it against the May 2026 generic-entry settlements.
2. If you do file, treat the § 315(b) one-year bar and discretionary-denial risk as gating items. The asserted MDL is in Delaware, parallel proceedings on the '708 are fully resolved (which cuts both ways on Fintiv-style efficiency analysis), and the statutory one-year trial deadline from institution, extendable by up to six months for good cause under § 316(a)(11), will run into the patent's expiry — a strong argument for denial.
3. Ground your invalidity case in the '921's own prosecution history, not the '708 IPR. The examiner's surfactant/lubricant rejections (the '922, '578, '752, and '810 references) and applicants' § 1.132 declaration are the sharpest available hooks, and none of them are estopped.
4. Assess issue preclusion, not just estoppel. If you are in privity with Mylan (e.g., a co-defendant in the same MDL or a supply-chain partner), the 2022 judgment that '921 claim 1 is not invalid may be asserted against you collaterally. Map your privity chain before you assume you are a clean slate.
5. Verify against primary sources before relying on this memo. I could not exhaustively query PTAB E2E / PTAB Decisions. Confirm the zero-proceedings result directly at PTAB E2E, and pull the '708 FWD from that same system. Key documents I did verify:
   • N.D. W. Va. bench trial opinion (Dkt. 204): https://www.courtlistener.com/recap/gov.uscourts.wvnd.46385/gov.uscourts.wvnd.46385.204.0.pdf
   • Fed. Cir. Mylan v. Merck, No. 2021-2121: https://www.mololamken.com/assets/htmldocuments/Mylan%20v.%20Merck%20Opinion.pdf
   • MDL Markman opinion (D. Del. 2020-11-17, '708 & '921), Case 1:19-cv-00311-RGA Dkt. 135

---

Caveats and confidence. High confidence that no AIA trial proceeding has been instituted on the '921 patent — both the canonical ODP block and my search results are consistent, and I found no contrary indicator. Moderate confidence in the completeness of the family-related list (IPR2020-00040, -01045, -01060, -01072); I did not locate a separate Viatris/Teva appeal docket number and have not supplied one. I have not stated any claim-level outcome for the '921 patent, because none exists. Where dates appear in two forms in the sources (e.g., the '921 bench ruling announced 2022-09-22 versus the sealed opinion docketed 2022-10-26; the IPR2020-00040 petition filed 2019-10-30 under a "2020" case number), I have flagged the discrepancy rather than reconciled it silently.

I'll verify the assignment record and litigation context before writing.

Let me try to pull the actual USPTO assignment records with reel/frame data.

Assignment & Ownership Analysis — US 8,414,921 B2

Data-source note: The authoritative patent text supplied reproduces the USPTO assignment events (recorded dates, conveyance types, assignors/assignees) as mirrored in Google Patents legal events, but it does not include reel/frame numbers or correspondent-of-record fields. Those two fields require a direct record-by-record pull from the Assignment Center, which I was unable to complete from the sources available; I flag every resulting gap below rather than inventing values. Nothing below is speculative — each link is a recorded corporate action.

---

Inventors

All five are named as assignors on the 2008-05-29 recorded assignment to Merck & Co., Inc., confirming a single-employer inventor group (Merck's pharmaceutical formulation/commercialization team). No unusual patterns: there is no record evidence of inventors departing the original assignee, no separate inventor-to-third-party conveyances, and no inventor co-ownership. (Whether individual inventors left Merck within 12 months is not determinable from the assignment record — that data point is not present in the available sources.)

Original assignee

Merck & Co., Inc. (subsequently renamed Merck Sharp & Dohme Corp., now Merck Sharp & Dohme LLC).

• Product practice: Yes — the claims read on JANUMET® (sitagliptin phosphate + metformin HCl fixed-dose tablet), FDA-approved in 2007, plus JANUMET® XR. The patent is listed as a JANUMET patent and is the last-expiring Orange Book PED-reflected reference for the product (adjusted expiration 2028-07-21; pediatric-exclusivity date cited as 2029-01-21). This is a blockbuster commercial product, not a paper asset.
• Primary line of business: Research-based pharmaceutical manufacturer (vaccines, branded medicines, biologics).
• Current status: Operating. Publicly traded (NYSE: MRK); the 2022 assignee (Merck Sharp & Dohme LLC) is the current Rabway, NJ operating subsidiary. No bankruptcy, wind-down, or asset-sale proceedings.

Assignment timeline

The Assignment Center contains recorded assignments for this patent (five events). Retrieved dates and conveyances are from the USPTO-mirrored legal events; reel/frame values were not retrievable and are marked as such.

• Executed on/about filing (2006) / recorded 2008-05-29 — Reel not retrieved

  • Conveyance: Assignment
  • Assignor: Kamali, Alani, Fliszar, Ghosh, Tijerina (individual inventors)
  • Assignee: Merck & Co., Inc.
  • Correspondent: not retrieved (gap — requires Assignment Center pull)
  • Context: Standard employee inventor-to-employer assignment perfecting Merck's title in the application.

• Executed n.d. / recorded 2010-01-26 — Reel not retrieved

  • Conveyance: Change of Name
  • Assignor: Merck & Co., Inc.
  • Assignee: Merck Sharp & Dohme Corp.
  • Correspondent: not retrieved
  • Context: Internal corporate renaming — no change in beneficial ownership.

• Executed n.d. / recorded 2012-08-27 — Reel not retrieved

  • Conveyance: Merger
  • Assignor: Merck Sharp & Dohme Corp.
  • Assignee: Schering Corporation
  • Correspondent: not retrieved
  • Context: Step in the Merck / Schering-Plough combination (Schering-Plough survived as the legal vehicle) — internal reorg, not a transfer to a third party.

• Executed n.d. / recorded 2012-08-29 — Reel not retrieved

  • Conveyance: Change of Name
  • Assignor: Schering Corporation
  • Assignee: Merck Sharp & Dohme Corp.
  • Correspondent: not retrieved
  • Context: Surviving merger vehicle renamed back to Merck Sharp & Dohme Corp. — completes the 2012 reorg.

• Executed 2022 (deed/certificate of merger dated 2022-04-07) / recorded 2022-08-08 — Reel not retrieved

  • Conveyance: Merger
  • Assignor: Merck Sharp & Dohme Corp.
  • Assignee: Merck Sharp & Dohme LLC
  • Correspondent: not retrieved
  • Context: Final conversion of the corporate entity form (Corp. → LLC) within the Merck family; corroborated by international registers (effective ~2022-09-22).

Every recorded conveyance is either (a) the original inventor-to-employer assignment or (b) an intra-Merck name change/merger. There is no assignment to any third party, no security interest, no license recorded against title, and no release.

Timeline diagram

timeline
    title Ownership of US 8414921
    2006 : PCT application filed by Merck
    2008 : Inventors assign to Merck and Co Inc
    2010 : Name change to Merck Sharp and Dohme Corp
    2012 : Merger into Schering Corporation
         : Name change to Merck Sharp and Dohme Corp
    2013 : Patent granted
    2019 : ANDA litigation filed in Delaware
    2022 : Merger into Merck Sharp and Dohme LLC

NPE / troll-pattern signals

1. Shell-entity transfer — not present. All assignees are Merck family operating entities (Merck & Co., Inc. / Merck Sharp & Dohme Corp. / Merck Sharp & Dohme LLC); none carries an "IP/Holdings/Ventures/Licensing" suffix, none is a single-purpose Delaware or Texas LLC formed to hold the patent, and no products-absent licensee appears in the chain.

2. Known asserter in the chain — not present. No assignee matches Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Conversant/Mosaid, Vringo, Pendrell, Round Rock, or any Unified Patents / RPX high-frequency plaintiff list. The current owner is a Fortune-100 branded pharmaceutical company.

3. Repeat correspondent across the chain — unclear / data gap. Correspondents of record were not retrievable from the sources available, so this signal cannot be scored either way. Note that even if a single firm filed all five recordings, that would be expected for a large corporate patent department (e.g., the internal patent counsel signature block pattern) and, standing alone, is not an NPE finding. No recurrence evidence exists to flag.

4. Cascading transfers — not present (superficially similar, materially different). There are consecutive recordings in 2012 (Aug 27 merger, Aug 29 name change) and a further merger in 2022, which superficially resembles chained transfers; however, the chain is confined to one corporate family across 16 years and reflects the Merck/Schering-Plough merger and a Corp.→LLC conversion, not LLC-to-LLC hops through shell vehicles. No shared registered-agent address or third-party principal appears.

5. Pre-litigation transfer — not present. The asserted suits are the 2019–2020 actions (D. Del. 1:19-cv-00315/00316/00317/00319/00347, 1:19-cv-01489, 1:20-cv-01099; N.D. W. Va. 1:19-cv-00101). The nearest assignment is the 2022-08-08 merger, which post-dates the suits by ~2 years and is a name-change reorg, not an assertion-enabling transfer.

6. Bankruptcy fire-sale — not present. No Chapter 7/11 filing by any assignor; Merck is solvent and operating.

7. Privateering — not present. No transfer to an NPE asserting on Merck's behalf; Merck litigates in its own name and, per the collateral corporate records, licenses the sitagliptin/metformin technology through ordinary commercial agreements.

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

Verdict

Operating-company assertion.

The ownership chain begins with the inventors' assignment to Merck & Co., Inc. (recorded 2008-05-29) and consists solely of intra-Merck name changes and mergers — Merck Sharp & Dohme Corp. (2010-01-26 change of name), Schering Corporation (2012-08-27 merger), back to Merck Sharp & Dohme Corp. (2012-08-29 change of name), and finally Merck Sharp & Dohme LLC (recorded 2022-08-08) — with no third-party, shell-entity, or asserter ever entering the chain. The patent is a live Orange Book reference for the Merck-marketed JANUMET® product (adjusted expiry 2028-07-21) and is being asserted by Merck against ANDA filers in its own name in the 2019–2020 Delaware and West Virginia suits; that is textbook operating-company assertion, not NPE conduct. The only open data gap is correspondent-of-record (and reel/frame), which does not change the verdict.

Verify: USPTO Assignment Center — search patent number 8414921; mirror at https://assignment.uspto.gov/patent/index.html. Chain corroboration: Google Patents legal events for US8414921B2.

I'll research the specific prior-art references cited against US 8,414,921, verifying each identifier literally.

Let me gather details on the remaining key cited references.

I've hit the tool-call limit, so I'll now synthesize the analysis from the authoritative full patent text (which contains the complete citation lists with dates) plus the search results I successfully retrieved. Where I could not independently verify a reference's content, I say so.

---

Prior-Art Analysis — US 8,414,921 B2

0. Scope, legal framework, and a critical caveat

Applicable law. US 8,414,921 issued from application 12/085,722, filed 12 Dec 2006 (national stage of PCT/US2006/047380), claiming priority to 16 Dec 2005. Because the effective filing date predates the AIA's 16 Mar 2013 change, the pre-AIA version of 35 U.S.C. § 102 governs. The critical dates are therefore:

• § 102(b) statutory bar: publications/patents dated before 12 Dec 2005 (one year before the 12 Dec 2006 filing).
• § 102(a): art published/known before the invention date (presumptively 16 Dec 2005).
• § 102(e): US patents and US application publications (and PCT publications designating the US) having an effective US filing date before the invention date, even if published later.

A caveat I must state plainly. The task asks me to assign each cited reference to the claim(s) it "potentially anticipates." After mapping the references against the actual claim language, the honest conclusion is that none of the cited references fully anticipates the asserted claims — this is precisely why the patent survived the 2019–2022 litigation (per the prior section, Mylan did not even contest infringement of claim 1). The cited art is overwhelmingly § 103 obviousness art, not § 102 anticipation art. I set out below which references could on a stretch be argued to anticipate, and the specific missing element for each. I flag invention-date assumptions explicitly rather than overstating.

What the claims actually require (the "anticipation test" they must satisfy):

The two "fingerprint" limitations that separate claim 1 from the generic DPP-4/metformin idea are the narrow surfactant window (0.5–1%) and the mid-size diluent window (5–15%), coupled to sitagliptin specifically. A reference must disclose all of these to anticipate.

---

1. The cited references, ranked by relevance

I have grouped the 45 citations and 30 "Citations" entries from the Google Patents record by technological tier. Dates are the priority date / publication date as listed in the patent's own citation tables. I do not supply filing dates where the patent record does not give them and I could not independently verify them.

TIER 1 — DPP-4 inhibitor + metformin combination art (the closest art)

TIER 2 — Formulation / excipient / dosage-form art (relevant to the excipient limitations)

TIER 3 — DPP-4 inhibitor compound/method art (background, not formulation art)

Note on the "similar number" instruction: I confirmed that all items above are the references actually listed within the US 8,414,921 record (Google Patents "Citations (30)" and "Patent Citations (45)" tables). I did not substitute any adjacent-number document.

---

2. The non-patent literature cited (and its § 102 significance)

These are, in my view, the references most likely to be argued as § 102 art against the method claims 16/20, because they directly disclose sitagliptin (MK-0431) + metformin in Type 2 diabetics:

---

3. Bottom line — the § 102 anticipation assessment

No cited reference, alone, anticipates any claim of US 8,414,921. The specific reasons, claim-by-claim:

1. Claims 1, 21 (the composition claims). The closest references — WO 2001/052825, WO 2001/097808, WO 2006/047248, WO 2007/041053 — either (a) use a different DPP-4 inhibitor (vildagliptin in WO 2007/041053), or (b) disclose only a genus-level "DPP-4 inhibitor + metformin + carrier" combination without the limitation-critical surfactant 0.5–1% and diluent 5–15% windows, and without the specific lubricant/binder ranges. A reference cannot anticipate a composition claim by disclosing the two actives alone.

2. Claim 10 (unit-dose composition). No cited reference discloses sitagliptin at the recited unit strengths together with metformin at the recited FDA strengths and the ~1–2% lubricant / ~7% binder / ~10% diluent / ~0.5% surfactant ratios.

3. Claims 16 and 20 (methods). The method claims are composition-limited — they require administering "a pharmaceutical composition of claim 1" / "of claim 12." Brazg (2005) and Herman disclose co-administration of sitagliptin and metformin, which is not the fixed-dose composition of claim 1. So even the strongest § 102 candidate fails the composition limitation.

4. Claims 22–28 (product-by-formula). These recite the exact gram quantities from Examples 1–7 (e.g., claim 22: 64.25 mg sitagliptin phosphate monohydrate / 500 mg metformin HCl / 48.2 mg PVP / 3.45 mg SLS / 59.3 mg MCC / 13.8 mg sodium stearyl fumarate / 17.2 mg film coat). No cited reference discloses these specific completed formulas.

Consequence: The cited art functions in the file and in litigation as § 103 obviousness art (the combination concept — see Tier 1 — combined with the routine excipient art of Tier 2), not as § 102 anticipation art. This is consistent with the reported litigation posture in the earlier section, where Mylan did not contest infringement of claim 1 of the '921 patent.

---

4. Explicit confidence and limitation statement

• High confidence: the identity, priority dates, publication dates, assignees, and titles of all references — these are transcribed from the patent's own citation tables in the authoritative full text, and I independently corroborated the key ones (WO 2001/052825, WO 2001/097808, WO 2006/047248, WO 2007/041053, WO 1999/038501) against Google Patents/PubChem/Espacenet search results.
• Moderate confidence: the content characterization of references whose full text I could not open (notably US 2005/0051922 (Nangia), US 2006/0210627 (Pfeffer), and EP 1 557 165 (Eisai)). I have described them at the level their titles and abstracts support and flagged them as unverified; I would not represent their disclosure scope in a filing without pulling the full documents.
• Not established / would not rely on: whether any of the § 102(e)-candidate references (Tier-1 #3–4, Tier-2 #8, Tier-3 #23–24, #26) were actually applied by the examiner under § 102(e); the patent record supplied here shows a citation relationship, not the examiner's art rejection.
• Rule compliance: I interpreted every alphanumeric identifier literally (e.g., the specification's "WO 2005/0031335" is reproduced as written, not corrected to WO 2005/003135; the "anhydrite" typographical variants in claims 25/27/28 are noted in the earlier section and not silently fixed). No similar-but-different patent number was substituted for 8,414,921.

Note on the current date: the task header states 26 April 2026 while the system date is 28 September 2026. I flag this discrepancy rather than reconcile it, consistent with the operating rules; it does not affect the prior-art analysis, whose critical dates are all in 1998–2008.

I'll ground this obviousness analysis in the prior-art references actually cited on the patent page, and check what the courts/PTAB have said about §103 for this patent.

Critical finding: Mylan only challenged the '708 patent (OTDP + §112) at trial — not the '921 under §103. Let me verify the appellate outcome and dig into the specific prior-art references.

Obviousness Analysis — U.S. Patent 8,414,921 under 35 U.S.C. § 103

Date note: this task specifies April 26, 2026 as the analysis date; the system clock reads September 28, 2026. I use the task date below and flag that the two do not match. All statements build on the earlier Patent summary, Litigation summary, and Assignment sections rather than repeating them.

---

0. Two corrections/additions to the earlier sections — flagged up front

A. Material refinement to the Litigation section. The earlier summary said the Mylan action "held both patents valid and infringed." That is right as to outcome, but the scope of the validity challenge was narrower than that phrasing implies. Judge Keeley framed the trial issues as:

"(1) do Mylan's ANDA products infringe claim 3 of the '708 patent or claim 1 of the '921 patent; (2) are claims 1, 2, 3, and 19 of the '708 patent invalid under the judicially created obviousness-type double patenting doctrine; and (3) are claims 1, 2, 3, and 19 of the '708 patent invalid under 35 U.S.C. § 112…"
— D.I. 204 at 1–2, https://storage.courtlistener.com/recap/gov.uscourts.wvnd.46385/gov.uscourts.wvnd.46385.204.0.pdf

There was no § 103 challenge to the '921 patent tried in N.D. W. Va. The '921 was the infringement patent; validity was contested only against the '708. This matters because it means there is no merits adjudication of the '921's obviousness, in either direction, from that case.

B. New docket material not identified in the earlier sections. The Federal Circuit appeal from that judgment exists and is captioned Merck Sharp & Dohme, LLC v. Mylan Pharmaceuticals Inc., No. 2023-1013 (Fed. Cir.), originating from 1:19-cv-00101-IMK, notice of appeal docketed 10/19/2022 (https://www.docketalarm.com/cases/US_Court_of_Appeals_Federal_Circuit/23-1013/Merck_Sharp_%26_Dohme_LLC_v._Mylan_Pharmaceuticals_Inc/docs/3.pdf). The earlier section could not confirm an appellate docket; this supplies it. I did not retrieve a merits disposition and therefore say nothing about the outcome. (This is distinct from No. 2021-2121, Mylan's appeal of the '708 IPR — PTAB IPR2019-01387 — which the Federal Circuit affirmed on substantial-evidence grounds; see https://www.mololamken.com/assets/htmldocuments/Mylan%20v.%20Merck%20Opinion.pdf.)

C. A proceeding the earlier sections missed. The '921 patent was the subject of a PTAB proceeding captioned Mylan Pharmaceuticals Inc. v. Merck Sharp & Dohme Corp., IPR2020-00040, whose Patent Owner exhibit (filed 12/30/2020) reproduces the '921 prosecution history on the term "surfactant" for "('921 pat. claims 1, 3, 5–8, 10, 11, and 21)":
https://www.docketalarm.com/cases/PTAB/IPR2020-00040/Mylan_Pharmaceuticals_Inc._v._Merck_Sharp_%26_Dohme_Corp/docs/12-30-2020-Patent_Owner/Exhibit-2284-EX2284___Request_for_Certificate_of_Correction.pdf
I could not confirm institution or disposition from the retrieved material. Treat this as a lead to verify on PTAB docket 2020-00040, not as an established fact.

---

1. Legal framework, the patent's own admissions, and the level of ordinary skill

1.1 Which statute

The '921 has an effective filing date of 16 December 2005 and was filed 12 December 2006 — before the 16 March 2013 AIA change. Pre-AIA § 103(a) governs. That matters here for two reasons: (i) the pre-AIA § 103(c)(1) common-ownership safe harbor exists and can disqualify certain prior art; and (ii) pre-AIA § 102(e)/(g) reference dates control the availability of published applications.

1.2 The patent's own specification is a Graham-factor-1 problem for Merck

A substantial amount of the motivation a challenger needs is written into the patent itself:

• "Combining two or more oral antidiabetic agents into a single tablet provides a potential means of delivering combination therapy without adding to the complexity of patients' daily regimens." (col. 1)
• "Such formulations have been well accepted in other disease indications… Examples of marketed combination tablets containing two oral antidiabetic agents include Glucovance™ (metformin and glyburide), Avandame™ (metformin and rosiglitazone), and Metaglip™ (metformin and glipizide)."
• "…it is essential that the components have complementary mechanisms of action and compatible pharmacokinetic profiles." (This is the patent reciting its own motivation-to-combine test.)

The specification then lists the entire excipient set as conventional: "diluents, compression aids, glidants, disintegrants, lubricants, flavors… known in the pharmaceutical formulation art." Claims 1's ingredients are therefore admitted to be a selection among known excipient classes — which pushes the whole § 103 fight onto (i) motivation to make the sitagliptin/metformin FDC at all, and (ii) whether the six numeric windows are routine optimization.

1.3 Level of ordinary skill (POSITA)

Most plausibly: a B.S./M.S. in pharmacy, pharmaceutical chemistry, or chemical engineering with ~3–5 years, or a Ph.D. with ~1–2 years, in oral solid-dosage development, including wet-granulation and roller-compaction experience with high-drug-load tablets. The '921 specification is itself written at roughly that level, and the W. Va. opinion's framing ("The invention in claim 1 is a pharmaceutical composition comprised of sitagliptin, metformin, a lubricant, a binding agent, a surfactant, and a diluent, each within a specific weight range") is consistent with a formulation-scientist POSITA.

---

2. The prior art actually of record, with availability dates

Drawn from the "Citations (30) / Patent Citations (45) / Non-Patent Citations (5)" blocks on the patent page, plus the Novartis family material retrieved.

---

3. Claim 1 element-by-element, mapped to the art

Key structural observation: with (d) at "0–35%" and (c) at "0.1–10%," the limitations doing real work are (e) and (f) — precisely the two the W. Va. court found necessary to reach for infringement: "To establish infringement Merck must prove … that Mylan's product also satisfies limitations (e) and (f), related to the surfactant and diluent components" (D.I. 204 at § 3). A § 103 attack therefore has to succeed or fail on 0.5–1% surfactant and 5–15% diluent, in a sitagliptin/metformin tablet.

---

4. Proposed § 103 grounds

Ground 1 (strongest): Ahren/Brazg/Herman + WO 01/52825 (or WO 01/97808) + '922

Combination: Ahren '05 and/or Brazg '05 (clinical combination of a DPP-4i with metformin; Brazg names sitagliptin) in view of WO 01/52825 (Novartis: DPP-4i + antidiabetic combinations) and US 2005/0051922 ('922: a working metformin HCl tablet with PVP binder, SLS and magnesium stearate).

Motivation to combine — this is where the ground is strongest, because four independent sources converge:

1. Clinical necessity. The '921 specification itself concedes monotherapy "do[es] not sufficiently control glycaemia during long-term treatment, leading to a requirement for combination therapy within several years following diagnosis."
2. Clinical proof. Ahren '05 demonstrates one year of DPP-4i + metformin co-therapy improving β-cell function/insulin sensitivity; Brazg '05 does the identical experiment with sitagliptin (MK-0431) specifically. A POSITA reading Brazg has the exact drug pair in hand.
3. Express combination teaching. WO 01/52825 and WO 01/97808 are titled and directed to DPP-4i + antidiabetic combinations, metformin being the archetypal biguanide. The concept-level motivation is not inference; it is the references' stated purpose.
4. PK compatibility — the patent's own gate. The specification says the components must have "compatible pharmacokinetic profiles." Herman et al. (cited on the '921's own face) supplies exactly that for MK-0431 + metformin. A challenger argues that the patent's stated precondition for combining was already satisfied in the literature before the priority date.

Reasonable expectation of success on the formulation. Metformin is a known brittle, poorly compressible, high-dose API — '922 states it outright: "metformin formulations require a very high percentage of active ingredients (up to 1000 mg), which leaves minimal room for excipients… The ability to improve the compressibility… is generally limited to techniques such as wet granulation with a binder." '922 then supplies the binder (PVP), the surfactant (SLS, which the reference itself calls "a hydrophobic surfactant"), the lubricant (magnesium stearate), the strengths (500/1000 mg) and the granulation route. A POSITA would have a reasonable expectation of obtaining a compressible, immediate-release tablet.

KSR overlay: KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) — a predictable variation using known excipients for their known functions, driven by a known market demand for FDC antidiabetic tablets (the specification lists Glucovance/Avandamet/Metaglip as evidence of that demand), is the paradigm "combination of familiar elements according to known methods." In re Aller, 220 F.2d 454 (CCPA 1955), and In re Woodruff, 919 F.2d 1575 (Fed. Cir. 1990), supply the range-optimization doctrine: where the prior art discloses the parameter as result-effective, optimizing its value is routine. SLS level in a metformin granulation is such a parameter ('922's data plot hardness as a function of SLS percentage).

The "about" problem cuts against the patent. The W. Va. court treated "about 0.5 to 1%" as extending below 0.5% ("how far does the term 'about' extend the lower end…?"). Whatever the precise accused value, the practical scope of limitation (e) is wider than the literal window — which means '922's SLS levels (whether ~0.5%, ~2.5% or ~4.6% depending on how its internal percentages are read) land closer to the claim than the literal numbers suggest.

Ground 2: '922 + WO 01/52825 + general knowledge of FDC tablet practice + the specification's admissions

Same core, but the challenger leans on the specification's own admissions to supply the FDC-tablet motivation (the Glucovance/Avandamet/Metaglip passage + the "complementary mechanisms / compatible PK" passage) and on '871/'498 only for the identity of sitagliptin. This is the "predictable variation of a known product form" theory and is the ground most likely to be run in a petition, because it does not need a reference that literally co-formulates the two drugs.

Ground 3 (claim 9 and 22–28): everything above + '871/'498 + WO 2005/0031335

• Claim 9 (dihydrogenphosphate salt) is the '708 patent's subject matter. A challenger must contend with the Federal Circuit's affirmance in Mylan v. Merck (No. 2021-2121) that the DHP salt and its monohydrate were neither anticipated nor obvious over WO '498 + Brittain + Bastin. Claim 9 is the hardest claim in the '921 to invalidate, and its presence creates a fallback position even if the formulation claims fall.
• Claims 22–28 recite complete, closed tablet formulas. A challenger would argue these are the "commercial development" tables of the specification (about 9%/73%/7%/0.5%/10% etc.) and therefore product-by-formula claims reached by routine optimization of the Ground 1 combination. This is the most vulnerable family: nothing in the retrieved art teaches the specific mg-for-mg formula, so the attack is purely "optimization of a known formulation." Expect a strong non-obviousness rebuttal here (§ 5.4).

Ground 4 (method claims 16 and 20)

If the composition is obvious, the methods fall with it — using a DPP-4i + metformin to treat T2DM was squarely known (Ahren, Brazg, WO 01/52825, WO 01/97808). Claim 20's added limitation (sitagliptin 25/50/75/100/150/200 mg with metformin 500/850/1000 mg) adds nothing that the clinical literature did not already suggest — Brazg's protocol used sitagliptin with metformin at approved strengths.

---

5. Where the § 103 attack breaks down — and it breaks down in four places

5.1 No pre-16-Dec-2005 reference discloses the FDC tablet

This is the single biggest obstacle. Every retrieved pre-priority reference teaches one of: (i) co-administration of separate dosage forms (Ahren, Brazg, Herman); (ii) a genus-level combination claim (WO 01/52825, WO 01/97808); or (iii) a metformin-only tablet ('922). None discloses a single dosage form containing both sitagliptin and metformin. A challenger must therefore bridge from "co-administer" to "co-formulate," and must do so on a record where the patent's own drafter wrote that co-formulation is what the field needed — an admission that helps the challenger on motivation but does not by itself supply the claimed windows.

5.2 The closest formulation art is not prior art at all

WO 2007/041053 / EP 1 948 149 (Novartis, "Formulation comprising metformin and vildagliptin") is the reference that would make this case easy: its Table 3-2 (500/50 mg) shows a 605 mg core of 8.2 wt% DPP-4 inhibitor, 82.7 wt% metformin HCl, 8.2 wt% HPC binder, 0.9 wt% magnesium stearate, Opadry film coat, made by wet granulation; Table 3-4 (1000/50 mg) shows 4.3/86.3/8.5/0.9. Header data confirm priority 29 Sep 2005 but international filing 25 September 2006 (see https://data.epo.org/publication-server/rest/v1.0/publication-dates/20111116/patents/EP1948149NWB1/document.html).

Under pre-AIA § 102(e), a published PCT designating the US gets the international filing date, not the provisional's date. 25 Sep 2006 post-dates the '921's 16 Dec 2005 priority date, and its 12 Apr 2007 publication post-dates the '921's 12 Dec 2006 filing. On its face, WO 2007/041053 is not § 102(a), (b) or (e) prior art to the '921. That is a serious evidentiary loss for a challenger, because '053's binder (7.6–8.7 wt%, squarely inside claim 1's 0–35% and claim 3's 4–9%), lubricant (~0.9 wt%) and wet-granulation process map almost one-to-one onto the '921's "commercial development" embodiments. (Caveat: I have not run a full § 102(e) date calculation on any US national-stage counterpart of the Novartis application; confirm before relying on this.)

5.3 The § 103(c) safe harbor — available for some references, not for the published sitagliptin art

The earlier Litigation summary noted that in the '708 IPR the Board applied pre-AIA § 103(c)(1) to knock out the Merck-owned Edmondson/'498 reference as to most claims. A challenger to the '921 can be expected to test whether the same shield applies — and it largely does not:

• § 103(c)(1) disqualifies only subject matter that qualifies as prior art solely under § 102(e), (f) or (g). US 6,699,871 issued 2 March 2004 and WO 03/004498 published 16 January 2003 — both more than one year before the '921's 12 December 2006 filing. They are therefore § 102(b) art, outside the safe harbor, and cannot be disqualified by common ownership.
• The shield would be available for any reference that reaches only § 102(e) status (e.g., WO 2006/047248, if its international filing date is confirmed to precede 16 Dec 2005).

Net: the mechanism that saved Merck in the '708 IPR is materially weaker for the '921, because the sitagliptin-disclosing references are § 102(b). This is the opposite of what the earlier section's framing might lead one to expect, and worth stating plainly.

5.4 Objective evidence

• Unexpected results / no reasonable expectation of success. The specification's examples report a single 689–1300 mg monolithic tablet delivering immediate release of both APIs at metformin loads of 500–1000 mg, prepared by both high-shear and fluid-bed granulation, with acceptable compression and stability. A POSITA at the priority date faced a genuine technical problem: a brittle, poorly compressible, highly water-soluble drug at 77 wt% load, plus a low-dose second API, in one tablet. In re Soni / WBIP line of authority supports weighting this.
• Teaching away. '922's data show extra-granular SLS improved hardness up to ~0.5% and then slightly decreased it (26.8 kp at the 0.25% step vs 25.6 kp at the 0.50% step), and '922 warns that hydrophobic lubricants "interrupt[] bonding sites between the particles surfaces thereby weakening the tablet structure." A challenger's own primary formulation reference thus contains a mild directional signal away from the claimed 0.5–1% surfactant band — a Kwik Products-style argument, though a weak one.
• Prosecution-history estoppel-style evidence. Applicants distinguished the '922 rejection on the record: "in '922, the sodium lauryl sulfate is used as an extra-granular absorption/compression enhancer to increase the hardness of the tablet, and the '922 does not disclose sodium lauryl sulfate as a surfactant… In the present invention, a surfactant, such as sodium lauryl sulfate, works as a wetting agent to facilitate liquid ingress… without a surfactant… the dissolution and bioavailability of the sitagliptin and metformin pharmaceutical composition would be adversely impacted" (IPR2020-00040, Ex. 2284). Note the vulnerability: '922's own specification calls SLS "a hydrophobic surfactant," which undercuts the distinction; a challenger will hammer that inconsistency.
• Copying. 20+ ANDA filers across the 2019–2020 wave (per the Litigation section) is objective evidence of copying, though it must be tied by nexus to the claimed formulation rather than to sitagliptin's efficacy.
• Commercial success. Janumet's blockbuster status supports non-obviousness only if nexus to the formulation claim is shown; expect a challenger to argue the success flows from sitagliptin's clinical profile, not from the 0.5–1% SLS or 5–15% MCC windows.

5.5 The 2020 PTAB attack

Mylan's IPR2020-00040 targeted '921 claims reciting "surfactant" (claims 1, 3, 5–8, 10, 11, 21) — i.e., an attack aimed squarely at limitation (e), using the same '922 prosecution record. I could not determine whether that petition was instituted or denied, or whether an FWD issued. If it was instituted and Merck prevailed, that is strong evidentiary support for non-obviousness of the surfactant window; if it was denied institution, it proves nothing either way.

---

6. Claim-by-claim conclusions

---

7. Bottom line

The best § 103 case against the '921 is:

Brazg et al. (2005) in view of WO 01/52825 (or WO 01/97808), further in view of US 2005/0051922, and optionally US 6,699,871 / WO 03/004498 for the identity of sitagliptin — with Herman et al. (2005) supplying the PK-compatibility precondition the patent itself recites, and the '921 specification's own admissions supplying the FDC-tablet motivation.

That combination answers Graham factors 1–3 reasonably well: the drug pair is known, the class combination is expressly claimed in the art, the clinical benefit is demonstrated, the PK compatibility is documented, and a working metformin tablet with PVP binder, SLS and magnesium stearate — made by wet granulation — is disclosed with 500 and 1000 mg strengths. Under KSR and In re Aller, the numeric windows are vulnerable as optimization of result-effective variables.

But the ground has three hard problems: (1) no pre-priority reference puts sitagliptin and metformin in one tablet; (2) the reference that would put the formulation windows squarely in the prior art (WO 2007/041053) appears, on its face, not to qualify as prior art; and (3) the sitagliptin-specific salt claim (claim 9) is protected by an affirmed Federal Circuit non-obviousness holding on related subject matter. Add the '922 "extra-granular enhancer ≠ surfactant" prosecution distinction and the absence of any judicial § 103 finding against the '921, and I would expect a validity challenge litigated today to be closely contested — plausibly surviving on claims 1/3–14 with a materially higher risk on claims 15, 16, 17, 18, 20, and especially 22–28.

I want to be explicit about the limits of this opinion: it is a prior-art-based prediction, not a legal conclusion, and no court has adjudicated the '921's obviousness. The Mylan trial did not reach it.

---

8. Confidence and verification list

High confidence: pre-AIA § 103(a) applies; the '921 validity challenge at trial was limited to the '708 (verbatim from D.I. 204); the identity and dates of the cited references; the content of US 2005/0051922 and the prosecution characterization of it; the captions/dates of WO 2007/041053 / EP 1 948 149.

Moderate confidence: the availability dates assigned in § 2 (I have not run independent § 102(b)/102(e) date calculations on each reference); the mapping of '922's internal SLS percentages; the PHOSITA definition.

Explicitly unverified — do not rely on without checking:

1. Disposition of IPR2020-00040 (Mylan v. Merck), including whether the '921 was the challenged patent and whether institution was denied or an FWD issued.
2. Disposition of Fed. Cir. No. 2023-1013 — the docket exists; I found no merits opinion.
3. The precise international filing date of WO 2006/047248 (needed for its § 102(e) status).
4. Whether any US national-stage counterpart of WO 2007/041053 has a § 102(e) date earlier than 16 Dec 2005.
5. The published full text of WO 01/52825 and WO 01/97808 — my characterization rests on their titles, abstracts, and the '921's own citation of them, not on a full read.
6. The W. Va. opinion's specific accused SLS and MCC values underlying its "about" analysis.

Suggested verification: PTAB docket IPR2020-00040; Fed. Cir. docket 2023-1013 on PACER; EPO Register for EP 1 948 149 to confirm the international filing date chain; USPTO Patent Center "Application Data" → prosecution history for 12/085,722 (the '922 rejection and the SLS-as-surfactant remarks).