Critical Analysis: Intra-Cellular Therapies, Inc. v. Controller of Patents

 

Critical Analysis: Intra-Cellular Therapies, Inc. v. Controller of Patents (2026:DHC:5394) C.A.(COMM.IPD-PAT) 24/2023 | Delhi High Court | Decided: 06.07.2026

I. Doctrinal Analysis: Novelty and the "Coverage vs. Disclosure" Question

A. The genus-species anticipation problem

The core novelty dispute was a classic Markush-genus-versus-species-selection issue. The appellant argued that arriving at the claimed species from the generic Formula I of D1/D7 required "multiple selections" among independent variables (R1–R6), and that the Controller impermissibly relied on more than one prior art document to construct a single "closest prior art" novelty attack — a submission with real doctrinal pedigree, since novelty (unlike obviousness) is ordinarily tested against a single prior document read as a whole.

The Court's response — invoking AstraZeneca AB and Boehringer Ingelheim v. Vee Excel — collapses the "covered vs. disclosed" distinction, holding that specific disclosure is immaterial where a compound falls within the claim scope of a genus patent. Before assessing that holding, one point of accuracy about the judgment should be registered: the Court did not mosaic D1 and D7 into a single novelty attack. It matched each claim family to one anticipating document — D7 to claims 1-3 and 5-10 (Formula I-III), and D1 to claims 4 and 5-7 (Formula IV) — so the single-document rule for anticipation was respected on the Court's own structure, and the appellant's "multiple prior arts as closest prior art" complaint is largely answered by that mapping. The genuinely live question is therefore not mosaicing across documents but whether the multiple selections required within a single genus (D1 or D7) defeat its anticipatory effect. On that question, the distinction between mere coverage and individualised disclosure does real work:

1. Infringement context: A generically-worded claim covers a species even if unexemplified (this protects the patentee).
2. Anticipation context: Whether an unnamed species falling within a prior generic formula is "disclosed" for Section 2(1)(j) purposes requires asking whether the skilled person, reading the document, would at once envisage that specific compound — the "photograph" test from General Tire-type genus-species jurisprudence, echoed in Indian precedent as well.

The judgment does not engage this classical genus-species framework (there is no reference to General Tire & Rubber Co. v. Firestone, or to the Indian decisions that apply a "reverse infringement" test with caution). Instead, it applies the coverage doctrine directly, treating a compound's falling within the claim scope of D1/D7 as equivalent to its being anticipated by D1/D7, and bypassing the "immediate and unambiguous disclosure" / "individualisation" requirement that novelty jurisprudence ordinarily demands for selection inventions. The objection is not that the doctrine is a misplaced transplant — it echoes infringement-side reasoning but was itself developed by this Court in the validity/anticipation register, in AstraZeneca and Boehringer, to stop a patentee defending a later species patent by pointing to the absence of express disclosure in its own earlier genus patent, and it is by now a settled feature of Indian validity analysis. The objection is rather that the Court reaches the result without confronting the individualisation question — and without naming the one doctrinal device that could justify the shortcut on these facts.

That device is the "person in the know". The respondent argued — relying on AstraZeneca — that where the inventor of the prior art and of the claimed invention are the same party, anticipation by publication is tested not against the ordinary "person skilled in the art" but against the "person in the know". It is that heightened-knowledge standard, and not the coverage doctrine alone, that allows same-applicant genus disclosure to be treated as anticipatory without any individualisation analysis. The persuasive answer to the Court is therefore not merely to invoke General Tire, but to contest the reach of the "person in the know" test: it was fashioned for anticipation-by-prior-claiming as between two granted patents in infringement and revocation litigation, and its migration into the ordinary examination of a pending application — where the applicant is not manipulating the coverage/disclosure line to escape its own grant, but resisting a novelty objection against a fresh species — is neither obvious nor self-justifying. A rigorous engagement with the decision must meet the "person in the know" standard head-on rather than assume the classical skilled-person test governs.

B. Comparative note

Under EPO practice (T 12/81, T 198/84 and the "purposive selection" line of case law), a selection from a broad prior art class is novel if it is not individualized in the prior document (the "two lists" principle governing selection from substituent lists such as R1–R6). Under current EPO practice, the technical-effect element — precisely the appellant's argument based on differential metabolic behaviour across R4/R5 positions — bears on inventive step, not novelty, following the 2019 revision of the Guidelines and T 261/15; the India/EPO divergence here may therefore lie as much in inventive step as in novelty. Tellingly, the appellant highlighted that the EPO's own Search Opinion, examining the same D1–D7 documents, found claims 1–16 novel. The Court did not meaningfully engage this comparative divergence beyond the (correct, but insufficient) observation that patent rights are territorial (per Communication Components Antenna v. Ace Technologies). Territoriality answers the question of whether foreign grants bind the Controller; it does not answer the separate question of why the same documents, evaluated under a materially similar "person skilled in the art" novelty standard, yielded opposite conclusions. A rigorous judgment would have distinguished the EPO's selection-invention framework on its merits rather than resting on jurisdictional non-bindingness alone.

II. Section 3(d): Correct Framework, but Compressed Application

A. Doctrinal fidelity

On this ground, the judgment is more solidly reasoned. It correctly extracts the Novartis triad — (i) identify the known substance and known efficacy; (ii) explain how the claimed substance is a derivative/new form; (iii) objectively compare therapeutic efficacy — and applies the Taiho Pharmaceutical three-factor test cited by the appellant itself. The Court's core holding — that improved pharmacokinetic/bioavailability data (reduced amide-metabolite formation, altered AUC ratios) is not, without more, "therapeutic efficacy" — tracks Novartis and the Division Bench's reasoning in Natco Pharma v. Novartis (2024) faithfully. This is legally sound and consistent with the settled principle that Section 3(d) demands a narrow, strict efficacy showing for known-substance derivatives.

B. The gap: identifying the "known substance"

However, the appellant's procedural objection — that neither hearing notice clearly identified the "known substance" as required by Taiho factor (i) — receives no direct answer. The Court's own analysis effectively does the Controller's work retrospectively, identifying compound Q as the known substance from D1/D7 paragraphs and then testing efficacy against it. This raises a natural-justice concern: if the Controller's order was silent or inadequate on identifying the known substance (as Taiho itself demands as a Controller obligation), is it appropriate for the appellate court to supply that missing step de novo rather than remanding? The judgment does not explain why curing this deficiency in appeal, rather than remand, was appropriate — particularly given that Section 117A appeals are not strictly limited to record review but the Controller's reasoning obligation is itself a natural-justice safeguard the Court elsewhere (implicitly) values. That said, the force of this objection is blunted by recording compound Q as an "admittedly" known compound and in substance the applicant's own parent molecule, around which the entire specification is built; the "known substance" was therefore not genuinely in dispute, and any failure to spell it out in the hearing notices carries a strong harmless-error character.

III. The Co-Inventor Affidavit and Natural Justice

This is the judgment's most doctrinally interesting — and most quietly resolved — issue. The appellant's strongest procedural argument was that the Controller's order made no reference whatsoever to Dr. Peng Li's affidavit, filed on three separate occasions, invoking Milliken & Co. v. Controller and Regents of University of California v. Union of India — both cases where failure to consider evidence on record was treated as a serious defect.

The Court's resolution turns on a characterisation question the analysis must confront squarely. A Section 117A appeal is a full merits appeal, not a narrow judicial review: the High Court is entitled — arguably obliged — to examine evidence the Controller overlooked, and it does exactly that, taking up Dr. Peng Li's affidavit and finding its data duplicative of the Example 7 material already held insufficient. On that (correct) view of the appeal, the Court's independent examination is not itself a defect; it is what a rehearing requires. Two residual points remain, both narrower than the original critique:

1. Whether reasons-giving is a free-standing right: The remaining question is whether the Controller's total silence on an affidavit filed three times is a defect that survives even a merits rehearing — that is, whether Milliken and Regents of the University of California treat the duty to engage with filed evidence as an independent natural-justice entitlement (remand-worthy in itself) or merely as an input into a result the appellate court may reach for itself. The judgment assumes the latter without argument. If Milliken means the former, curing the silence on appeal — however sound the ultimate conclusion — does not discharge the duty, because a speaking-order obligation is not satisfied by a reviewing court supplying the reasons after the fact.
2. The constructive path not taken: A more principled disposal would have either (a) remanded for the Controller to engage with the affidavit and give reasons, consistent with Milliken, or (b) expressly articulated a harmless-error doctrine for patent prosecution — explaining why a first-instance failure to consider evidence is cured where the appellate court, on a full rehearing, can assess the same technical data and finds it unavailing. Indian jurisprudence has not yet clearly developed the latter, and the judgment would have been stronger for naming and adopting it than for proceeding as though the question did not arise.

IV. The Inventive Step Non-Decision

The terse disposal — declining to examine Section 2(1)(ja) because novelty and 3(d) were upheld — is logically defensible: a claim anticipated by D1/D7 cannot, a fortiori, possess an inventive step over the same art, so a finding of anticipation subsumes obviousness. It nevertheless remains analytically unsatisfying in three narrower respects:

• Structural asymmetry with the appellant's own submissions: The appellant's strongest doctrinal argument was that the Controller's obviousness reasoning was "cryptic," relying on Hoffmann-La Roche v. Cipla and Agriboard for the proposition that motivation-to-combine must be articulated, not asserted. This is a well-established requirement in Indian obviousness jurisprudence (echoing TSM-type reasoning obligations). By declining to rule, the Court leaves unresolved whether the Controller's "D4-D6 discloses the principle... therefore obvious" reasoning meets this bar — a live issue of precedential value for future applicants facing similarly conclusory obviousness rejections.
• Precedential thinness: Litigants and the Patent Office are left without guidance on whether "generic disclosure of a principle + independent disclosure of the base compound = per se obviousness" is an acceptable analytical shortcut, or whether Hoffmann-La Roche's demand for an articulated "route" from prior art to invention still applies with force.
• Overlap with novelty reasoning risk: The appellant specifically flagged that the Controller conflated novelty/obviousness criteria with the Section 3(d) analysis. The Court's own approach — deciding 3(d) using data explicitly generated to counter obviousness/novelty objections (Example 7, the affidavit) — arguably repeats this same conflation without expressly addressing the appellant's objection on the point.

V. Practical Implications

For patent applicants (particularly in deuterated/isotope and selection-invention spaces):

• The judgment reinforces that pharmacokinetic superiority (improved half-life, reduced metabolite formation, altered AUC) will not, by itself, satisfy Section 3(d) — applicants must generate and specifically plead in vivo efficacy/pharmacodynamic outcome data (disease-model or clinical-surrogate endpoints), not merely exposure or metabolic-stability data, however statistically robust.
• Applicants relying on genus-to-species selection arguments cannot expect Indian courts to import EPO-style "individualization" or "purposive selection" doctrine readily; the "coverage is disclosure" reasoning of Boehringer/AstraZeneca is likely to be applied expansively against species claims falling within earlier Markush claims, even where those earlier claims are the applicant's own prior filings (raising acute self-collision/obviousness-type double patenting risk for portfolio "evergreening" strategies via deuteration, chiral resolution, or salt/polymorph selection).
• Foreign prosecution outcomes (EPO/USPTO grants on the same prior art) are not binding before Indian tribunals and, on the strength of this judgment, are unlikely to move the needle absent independent India-specific argument — applicants should not expect harmonization arguments to gain traction, and should build India-specific efficacy dossiers rather than relying on foreign file wrappers.
• Affidavit and declaration evidence should be filed with express, itemized correlation to each statutory ground (novelty/obviousness/3(d)) — the judgment shows that generalized technical affidavits addressing "unexpected benefits" broadly may be read narrowly and found "duplicative" of specification examples already before the Controller.

For opponents/generic challengers:

• The decision is a favorable precedent for challenging deuterated-analogue patents and other incremental pharmaceutical patents, reinforcing the Novartis/Natco line that bioavailability ≠ efficacy, and providing an additional Delhi HC authority for the "coverage defeats novelty" principle usable in post-grant oppositions or revocation actions against later species patents where an applicant's own earlier genus filings exist.

For the Patent Office / policy makers:

• The judgment's willingness to cure the Controller's silence on the co-inventor's affidavit (rather than remand) may inadvertently lower the incentive for Controllers to issue fully reasoned speaking orders, since appellate courts may simply redo the technical assessment. This is in tension with the broader jurisprudential push (visible in Agriboard, Hoffmann-La Roche) toward requiring reasoned, non-conclusory Controller orders. Policy-level reform (e.g., IPAB-successor mechanisms, or Patent Office practice directions) could usefully mandate explicit engagement with all filed evidence as a matter of procedural regularity, reducing reliance on ad hoc appellate curing.
• The non-adjudication of inventive step leaves the "motivation to combine" reasoning standard for chemical/pharmaceutical obviousness somewhat underdeveloped at the Division/Single Judge level, an area ripe for a future reference or fuller treatment.

VI. Actionable Takeaways for Advisory Notes

1. For deuterated-compound and selection-invention filings in India: Build the specification and prosecution record around direct therapeutic/pharmacodynamic efficacy data, not merely PK/metabolic-stability comparisons; anticipate that PK superiority alone will not overcome Section 3(d).
2. Do not rely on "multiple selections needed" arguments to defeat novelty where the applicant's own earlier-filed genus patents (or related-party prior art) cover the claimed variable combinations — Indian courts currently treat claim coverage as equivalent to disclosure for anticipation purposes, a stricter standard than EPO selection-invention doctrine, and for same-applicant prior art may apply the heightened "person in the know" test rather than the ordinary skilled-person standard.
3. File affidavits and declarations early, and insist on express Controller engagement with them at the hearing stage; while courts may cure Controller silence on appeal, doing so removes an argument that would otherwise support remand, and the outcome-focused curing is not guaranteed to favor the applicant.
4. Do not treat foreign search opinions/office actions (EPO, USPTO) as persuasive authority before the Controller or Delhi HC on novelty/inventive step — cite them, if at all, only as comparative background, with heavy independent India-specific technical argument.
5. Preserve inventive-step arguments distinctly and insist on Controller engagement with "motivation to combine" under Hoffmann-La Roche/Agriboard; do not assume that failure to rebut novelty/3(d) forecloses a separate, focused obviousness challenge in any subsequent proceeding (divisional filings, oppositions to related patents), since this judgment leaves that standard unaddressed on the merits.
6. For portfolio strategy: Where a company holds prior Markush/genus patents, subsequent deuterated/optimized species filings face a live and now judicially reinforced risk of anticipation by the company's own earlier applications — freedom-to-operate and patentability clearances should map claim scope across the entire prior-filed family, not just third-party art.

VII. Conclusion

The judgment reaches a defensible result under Section 3(d) doctrine but does so through reasoning that is doctrinally uneven: it applies the Novartis/Taiho efficacy framework with appropriate rigor, while treating novelty through an expansive "coverage equals disclosure" lens that under-engages classical genus-species/selection-invention principles and comparative (EPO) divergence on the identical prior art. Its handling of the co-inventor affidavit issue, while pragmatically resolving the appeal, sidesteps a genuine natural-justice question about curing Controller silence on appeal rather than by remand. Practitioners should treat the decision as a useful, applicant-unfriendly precedent on Section 3(d) and novelty-by-coverage in the pharmaceutical/deuterated-compound space, while recognizing that its inventive-step non-decision and procedural shortcuts leave some doctrinal questions for future litigation to resolve more fully.