Section 3(d) of the Patents Act, 1970 turns on a word the statute does not define. The provision refuses patent status to a new form of a known substance that does not result in the enhancement of the known efficacy of that substance, and everything depends on what efficacy means, who must prove it, and against which substance the comparison runs. For eight years after the provision took its present shape, those questions had no authoritative answer. They received one on 1 April 2013, when a two judge bench of the Supreme Court decided Novartis AG v Union of India, reported at (2013) 6 SCC 1, and held that in the case of a medicine efficacy means therapeutic efficacy, judged strictly and narrowly, and established by research data rather than asserted by counsel.

The Court was conscious of the weight of error in both directions, and said so at the outset. A wrong decision in the appellant’s favour would grant a monopoly the law did not intend, over a medicine on which lives depended. A wrong decision against the appellant would deny a patent to an invention that deserved one, and would do so in a judgment that would govern every later application of the clause. That consciousness explains the judgment’s most distinctive feature: its care to state, at each step, the limits of what it was deciding, and to reserve the questions the record did not require it to answer. The case is usually summarised as a contest between a multinational company and the Indian generic industry over the price of a cancer drug, and that contest formed the background the Court itself acknowledged. The judgment, however, decides a question of statutory construction, and it decides it in a sequence that most summaries flatten. The Court first identified the known substance, and that identification, more than any argument about efficacy, is where the application was lost. It then explained why the properties the applicant relied on could not count toward efficacy at all. Only after those two steps did it reach the evidentiary holding for which the case is cited: that an increase in bioavailability, even if real, does not by itself establish enhanced therapeutic efficacy. This article takes the judgment in that order, sets out the provision in full before construing it, and states the two cases made about the provision, the originator case and the access case, at full strength and without adopting either.

The clause in full, because the article turns on its words

Section 3 of the Patents Act lists what are not inventions within the meaning of the Act. Clause (d), in the form given to it by the Patents (Amendment) Act, 2005, reads as follows. “The mere discovery of a new form of a known substance which does not result in the enhancement of the known efficacy of that substance or the mere discovery of any new property or new use for a known substance or of the mere use of a known process, machine or apparatus unless such known process results in a new product or employs at least one new reactant.” An Explanation follows, and the Explanation carries much of the operative weight: “For the purposes of this clause, salts, esters, ethers, polymorphs, metabolites, pure form, particle size, isomers, mixtures of isomers, complexes, combinations and other derivatives of known substance shall be considered to be the same substance, unless they differ significantly in properties with regard to efficacy.”

Four features of that text discipline everything that follows. The first is the word “mere”. The clause does not say that a new form of a known substance is not an invention. It says that the mere discovery of such a form, where the form does not enhance the known efficacy of the substance, is not an invention. A new form accompanied by demonstrated enhancement stands outside the clause. The second feature is that the clause has three limbs, not one. It addresses new forms of known substances, new properties or new uses of known substances, and the mere use of a known process, machine or apparatus, with a saving for a process that yields a new product or employs a new reactant. The third feature is the Explanation’s deeming rule. The listed derivatives are considered to be the same substance as the known substance, by force of the statute, unless they differ significantly in properties with regard to efficacy. A polymorph therefore begins its legal life as the same substance as its parent, and the burden of displacing that statutory identity lies on the applicant. The fourth feature is what the text does not contain. It does not define efficacy, it does not specify the evidence by which enhancement is to be shown, and it does not say in terms whether the comparison runs against the known substance in the abstract or against some identified predecessor. Each gap was litigated in Novartis.

The three limbs deserve separate attention, because only the first has generated a Supreme Court construction and the other two are administered daily without one. The first limb, new forms of known substances, is the pharmaceutical limb in practice. Its field of operation is the derivative: the salt chosen for manufacture, the polymorph chosen for the tablet, the ester, the isomer, the metabolite. Its question is comparative and evidentiary: the known substance works in a known way, and the form is patentable only if that working is enhanced. The second limb, new properties and new uses of known substances, is wider and quieter. A known substance found to treat a second condition, or to possess a property nobody had recorded, is a discovery of something that was always true of it, and the limb withholds invention status from the finding as such. It sits adjacent to Section 3(i), which excludes methods of treatment, so that the territory of second medical use claims in India is fenced on two sides, and applicants frame such claims, where they frame them at all, as products or processes rather than as uses. The third limb, the mere use of a known process, machine or apparatus, carries its own internal saving: a known process that results in a new product, or that employs at least one new reactant, escapes the exclusion. The saving shows the clause’s method in miniature. Throughout Section 3(d), the statute asks not whether the subject matter is dressed as something new, but whether anything substantively new has been produced: a better working substance, a new product, a new reactant. Where nothing substantively new is produced, the claim is a mere use or a mere discovery, and the adjective decides the case.

It is equally important to see what the clause does not do, because two of its supposed effects belong to other provisions. It does not regulate the term of any patent: a granted patent runs its twenty years from filing under Section 53 regardless of the clause, and the clause operates by refusing a second or follow-on right, not by shortening the first. And it does not create the access machinery of the Act. Compulsory licensing under Chapter XVI, government use, and the Bolar style exception in Section 107A are separate instruments with separate conditions, owned by other articles in this series. Section 3(d) works earlier and more bluntly than any of them: it denies that the follow-on subject matter is an invention at all.

The text quoted above is the amended text, and the amendment history is legible inside it. Before 2005 the clause had no new form limb and no Explanation: it was addressed to the mere discovery of any new property or new use for a known substance, and to the mere use of a known process, machine or apparatus. The 2005 amendment did three things at once. It inserted the new form limb at the head of the clause, bringing salts, polymorphs and their kin within the clause’s field for the first time in terms. It attached the efficacy condition to that limb, so that the new form is excluded only where it does not result in enhancement of the known efficacy. And it added the Explanation, whose deeming list converts the recurring categories of derivative into the same substance unless a significant efficacy linked difference is shown. Each element answered a feature of the litigation the legislature expected: the limb identified the strategy, the condition stated the test, and the Explanation allocated the burden of proof. When the Court later held that the provision sets up a second tier of qualifying standards, it was describing a machine the amendment had visibly assembled for the purpose.

The provision must also be placed correctly inside the Act’s architecture, because a persistent misreading treats it as the whole of Indian patentability law for pharmaceuticals. It is one clause in a list of exclusions. The definition of invention sits in Section 2(1)(j): a new product or process involving an inventive step and capable of industrial application. Section 2(1)(ja) defines inventive step as a feature of an invention that involves technical advance as compared to the existing knowledge or having economic significance or both and that makes the invention not obvious to a person skilled in the art. A pharmaceutical application must satisfy those definitions, and where it claims a new form of a known substance it must additionally survive Section 3(d). The judgment’s structural contribution, examined in a later section of this article, is the holding that the clause operates in addition to the general requirements and not instead of them, so that a claim may be new, may involve an inventive step, and may still fail.

Why the clause exists: the 2005 settlement in outline

The clause took its present form in 2005, and the circumstances of that year explain its stringency, though the detailed passage record belongs to the amendment article in this series and is held here at the level the judgment itself used. India had accepted the obligations of the TRIPS Agreement, which required product patents for pharmaceutical substances after a transition period that expired on 1 January 2005. The Patents Act, 1970 had barred product patents for medicines since its enactment, permitting only process patents, and an indigenous generic industry had been built on that foundation. The 2005 amendment deleted Section 5, the provision that had imposed the bar, and in the same exercise rewrote Section 3(d), adding the new form limb and the Explanation. The Supreme Court recorded that in the parliamentary debates the amendment in Section 3(d) was the only provision the government cited to allay the fears of opposition members concerning the abuses to which a product patent in medicines might be vulnerable. The clause was, in the Court’s words, meant especially to deal with chemical substances, and more particularly pharmaceutical products.

The judgment situates the clause within a longer statutory biography, which it recites because the clause’s meaning depends on the regime it qualifies. The Patents Act, 1970, enacted on the recommendations of the Ayyangar Committee, had made a deliberate developmental choice: for food and medicines, only process patents were available, so that an Indian manufacturer who found a different route to a known molecule could lawfully sell the molecule itself. Three decades of that regime produced the generic industry whose representatives appear in the case as opponents, an industry built on reverse engineering and process chemistry rather than on the discovery of new molecules. TRIPS required the restoration of product patents for pharmaceuticals after the transition, and the 2005 amendment performed the restoration, deleting Section 5. Section 3(d) in its amended form was the counterweight written into the same statute at the same moment. The pairing is the context the Court kept returning to: the amendment simultaneously opened the door to product patents and installed, at the threshold of the patent office, a filter addressed to the specific apprehension that the door would be used to patent the same medicine repeatedly in successive forms. The clause is therefore not an anomaly inside a liberalising statute. It is half of the statute’s settlement, and the Court construed it as such, holding the two halves together by insisting that the filter did not undo the opening.

The transitional machinery mattered to the facts of the case. Applications for pharmaceutical product patents filed between 1995 and 2005 had been held in a mailbox, to be examined when the law changed, and applicants could in the interim obtain exclusive marketing rights. The Glivec application was one of those mailbox applications, and the mailbox and the exclusive marketing right are the subject of the article on the Patents Amendment Act 2005 and TRIPS, which owns that machinery. What matters here is the consequence the Court drew from the timing. The application had been drafted in 1998 under a regime that asked no efficacy question. The efficacy question was then created by statute while the application waited, and the applicant attempted to answer it later, by affidavits filed during prosecution. Much of the litigation is explained by that single fact.

The statute as a whole, its definitions, its exclusions and its procedure, is the subject of the pillar article on the Patents Act 1970 explained, and this article does not re-teach it. The ground here is narrower: one clause, one judgment, and the body of practice the two have generated between them.

The legislative signals the Court used, and the ones it refused

A striking feature of the judgment is the disciplined way it handles legislative history, because Indian courts are often criticised, from opposite directions, for ignoring Parliament’s purposes or for substituting them for the text. The Court’s method in Novartis was to work from the text outward, admitting history for two purposes only: to identify the mischief the amendment addressed, and to choose between constructions the text would bear. Three signals were admitted. The first was the assurance recorded in the debates: when the product patent regime for medicines was restored in 2005, members feared repetitive patenting and price effects, and the government pointed to this clause, and to no other, as the safeguard. The Court treated that record as evidence of the clause’s purpose, not as a substitute for its words. The second signal was the direction of travel between versions. The clause had been amended more than once, and the 2005 exercise had made it, in the Court’s assessment, even more constrictive than the version it replaced, adding the new form limb and the Explanation’s deeming list. A provision deliberately tightened is to be read as tightening: the amendment’s history supported the strict and narrow construction the Court gave to efficacy, because a legislature that had meant to be permissive does not progressively narrow a text. The third signal was structural and internal to the Act: the placement of the clause in a section that declares what are not inventions, read with the definitions in Section 2, from which the Court derived the second tier analysis examined below.

Against those admitted signals stand the materials the Court refused to use as construction tools, and the refusals are as instructive. It recited the TRIPS Agreement and the Doha Declaration at length, as the setting in which Parliament had acted, and then set them aside: they explained why the law changed in 2005, but they could not decide what the changed law meant, and the fairness of the agreement was not in issue before it. The recital was not ornamental. By narrating the international framework in full, including the declaration’s statement that the agreement should be interpreted and implemented in a manner supportive of the right of members to protect public health and to promote access to medicines for all, the Court placed the Indian amendment inside a recognised international conversation about the balance of the patent system, and then decided the case on the domestic text. The international materials thus performed a legitimating function without performing a legal one: they showed the Court knew the terrain on which its construction would be judged abroad, and they were then put down. It declined to treat the price of Glivec as a legal datum: price entered the judgment only as part of the tribunal record it was reviewing, and the construction of the clause proceeds in the judgment without a single step that depends on what the drug cost. It declined, finally, to reason from the desirability of the outcome in either direction. The construction it adopted would have been the same construction, on its own logic, for a drug of modest importance and a company of modest size. That self denial is the reason the judgment has proved administrable: examiners and later courts were given a test built from the text, the debates’ assurance and the amendment’s direction, and not a licence to weigh access against innovation case by case. The weighing had been done, once, by the legislature, and the Court’s account of its own role never loses sight of that allocation.

The molecule before the claim: what was already known

The judgment begins not with the claim but with a patent granted two decades earlier, and the order is deliberate, because the Court’s central finding is that the substance at the heart of the dispute was already known from that earlier document.

A team led by Jurg Zimmermann at Ciba-Geigy, a predecessor of Novartis, had synthesised a family of derivatives of N-phenyl-2-pyrimidine-amine. One derivative, given the code CGP 57148 and later the international nonproprietary name imatinib, inhibited protein kinases involved in tumour growth and showed valuable anti-tumour properties. The family was the subject of a United States patent application, filed in its final form on 28 April 1994 and granted on 28 May 1996 as United States Patent No 5,521,184, referred to throughout the judgment as the Zimmermann patent, with a corresponding European patent. The Zimmermann specification described the compounds and, critically for what followed, described them together with their salts. It explained that compounds bearing a basic group may form acid addition salts with, among others, sulfonic acids, and it stated that references to the free compounds should be understood as including the corresponding salts where appropriate. It described the anti-tumour and pharmacological properties of the compounds, supported by test results, and imatinib appeared in it as one of thirty seven worked examples, example 21.

Imatinib was developed into a drug for chronic myeloid leukaemia and certain other tumours. It works by inhibiting the abnormal tyrosine kinase produced by the BCR-ABL fusion gene that drives that leukaemia, an early and celebrated example of a targeted cancer therapy, a medicine aimed at the molecular defect of the disease rather than at dividing cells generally. The form approved by the United States Food and Drug Administration in 2001 and marketed as Gleevec in the United States and Glivec elsewhere was imatinib mesylate, the methanesulfonic acid addition salt of imatinib, in a particular solid state, its beta crystalline form. A crystalline form, or polymorph, is a distinct arrangement of the same molecules in the solid state. Different polymorphs of one substance share a chemical identity and differ in physical behaviour: in how they flow as a powder, in how stable they are over time and temperature, and in how readily they absorb moisture. Novartis filed a Swiss application for the beta crystalline form in July 1997 and, on 17 July 1998, filed the corresponding application at the Chennai Patent Office, numbered 1602/MAS/1998, claiming the Swiss priority date. The application asserted three advantages for the beta form over the alpha crystalline form of the same salt: more beneficial flow properties, better thermodynamic stability, and lower hygroscopicity. It did not assert that the beta form treated cancer better than any other form. On the contrary, it stated that all the indicated inhibitory and pharmacological effects are also found with the free base. That sentence, written in 1998 for a different legal regime, returned fifteen years later as a load bearing admission.

The choice of comparator inside the application deserves a moment’s attention, because the Court returned to it repeatedly. The beta form’s advantages were asserted against the alpha crystalline form of the same salt, and the alpha form was not a stranger to the applicant: it was another crystalline state of imatinib mesylate, described in the applicant’s own earlier work. By framing the invention as an improvement on the alpha form, the application positioned the beta form within the family of the salt’s solid states, and claimed distinction from its nearest relative. That is the natural frame for a polymorph filing, and in most jurisdictions it is the frame that matters, since novelty and obviousness are assessed against the closest prior art. Under the amended Indian clause the frame was a trap. The statute’s comparator is not the nearest relative but the known substance, and the known substance, once identified as imatinib mesylate itself, stood behind both crystal forms indifferently. Superiority over a sibling form, even if fully proved, could not answer a test that asked about superiority over the parent substance, and the application’s framing had conceded the family relationship on which the identification of the parent depended.

Two pieces of pharmaceutical practice explain why the claim was framed as it was, and why claims of this kind are filed at all. The first is salt selection. Many drug molecules are weak bases or weak acids that crystallise poorly, dissolve poorly or degrade in storage in their free form, and converting the molecule into a salt is the standard industrial answer: the salt is chosen at the development stage, long before marketing, and the marketed product is almost invariably a salt rather than the free base or free acid. A salt is therefore rarely an afterthought or an embellishment. It is the form in which the medicine exists commercially. The second practice is polymorph screening. A crystalline substance may pack in several arrangements, and the arrangements differ in stability, solubility and behaviour on the production line; a form that converts to another form on the shelf, or that absorbs moisture and degrades, is a commercial hazard. Screening for the most stable polymorph, and characterising it thoroughly, is ordinary, expensive and necessary development work. Neither practice is contrived for litigation, and the judgment does not suggest otherwise. The legal significance of both is structural rather than moral: the work is real, the advantages are real, and after 2005 in India the work and the advantages must additionally produce an enhancement of therapeutic efficacy before they can support a patent. The clause does not ask whether the development was genuine. It asks what the development achieved for the patient.

Because the application was a mailbox application, it was not examined until the law changed. In the interval Novartis obtained exclusive marketing rights for Glivec on 10 November 2003 and enforced them against Indian generic manufacturers who were already selling versions of the drug. The enforcement mattered beyond the parties. Indian companies, among them the later opponents, had lawfully built imatinib products on the process patent regime and were supplying the medicine at a small fraction of the imported price, and the exclusive marketing right interrupted that supply years before any patent had been examined, let alone granted. The episode fixed the public character of the prosecution that followed: when the application came out of the mailbox in 2005, its refusal or grant was understood on all sides to decide not merely a private right but the terms on which a leukaemia medicine would be supplied in India, and the opponents who filed against it included the patients’ association whose members’ treatment depended on the answer. The price difference recorded in the proceedings was stark, and it is addressed in its proper place below. The legal point for the narrative is simpler: by the time the application came to be examined in 2005, the drug was on the market, rival versions were on the market, and the question whether a patent should issue at all had acquired a commercial and public significance far beyond the file.

Four decisions before the Supreme Court

The application was contested at every level the system provides, and the four decisions it produced frame the questions the Supreme Court had to answer. Five parties filed pre-grant oppositions: the Cancer Patients Aid Association and four generic manufacturers, Natco Pharma, Cipla, Ranbaxy Laboratories and Hetero Drugs. The oppositions put the application in issue on anticipation, obviousness, the priority claim and Section 3(d) itself.

Who opposed the Glivec application, and on what grounds?

The five pre-grant opponents were the Cancer Patients Aid Association, a patients’ organisation, and Natco Pharma, Cipla, Ranbaxy Laboratories and Hetero Drugs, generic manufacturers already selling imatinib products. They contended that the claim was anticipated by the Zimmermann patent, that it was obvious, and that the claimed form was barred by Section 3(d).

The opponents also attacked the priority claim, contending that the Swiss date of 18 July 1997 had been wrongly claimed for an application filed in India a year later. Their participation mattered structurally as well as substantively. Indian law permits any person to oppose a patent application before grant, and the Glivec prosecution became the demonstration case for that mechanism: the Section 3(d) question was fully argued at the first stage of the system, on evidence, by parties whose interest was precisely the interest the clause was designed to protect, years before a court saw the file.

The Assistant Controller of Patents and Designs refused the application on 25 January 2006. The refusal rested on four grounds: the claim was anticipated by the Zimmermann patent, it was obvious to a person skilled in the art, the priority date had been wrongly claimed, and its patentability was barred by Section 3(d). Novartis responded on two tracks. It challenged the Assistant Controller’s orders, and it filed writ petitions in the Madras High Court attacking Section 3(d) itself, contending that the provision was unconstitutional for vagueness and bad for non-compliance with TRIPS. The two tracks were then separated. After the Intellectual Property Appellate Board became functional, the appeals against the refusal were transferred to it under Section 117G, while the High Court retained and decided the constitutional challenge.

What did the Madras High Court decide in 2007, and what did it refuse to decide?

A Division Bench of the Madras High Court, in a judgment delivered on 6 August 2007, dismissed the challenge to the provision. It held that Section 3(d) does not violate Article 14, rejecting the argument that the undefined word efficacy gave examiners an unguided and therefore arbitrary discretion. It declined to pronounce on TRIPS compliance.

On the TRIPS question the Court held that conformity with the agreement was a matter for the dispute settlement machinery of the World Trade Organization and not for a municipal court, and Novartis did not appeal the decision. The High Court’s refusal on the TRIPS question is the first of two refusals that together define the limits of the case law, and its reasoning deserves precision. The Court did not hold that Section 3(d) complies with TRIPS. It held that the question was not the court’s to answer. A declaration of compliance or non-compliance from a domestic court would have bound no one internationally, and the treaty provides its own forum, in which one member state may complain against another before a panel. No member state has ever brought such a complaint against the provision, a fact whose significance is examined in the section on the two cases below. What the High Court did decide was the constitutional question, and its holding that an undefined statutory term, capable of being given content by the courts in the ordinary way, is not for that reason void for vagueness or arbitrary, was a conventional application of Article 14 doctrine. The provision survived, and the argument about its meaning moved to the forum where meaning is ordinarily settled, the appeal on the application itself.

What did the Intellectual Property Appellate Board hold, and what did it leave open?

The Appellate Board, by a detailed judgment dated 26 June 2009, reversed the Assistant Controller on anticipation and on obviousness, holding the claimed form novel and not obvious, and held Novartis entitled to the Swiss priority date of 18 July 1997. It refused the product patent all the same, on Section 3(d) alone.

Enhanced efficacy, the Board held, had not been shown, and that failure was enough. It allowed a process patent for the preparation of the beta crystalline form, the process claims standing outside the clause’s objection to the product. It added an observation, resting on the price charged during the exclusive marketing rights period, that a grant would also offend Section 3(b), the exclusion of inventions whose commercial exploitation prejudices public order, morality or public health.

The Board’s division of the case is the cleanest statement of the structural point in the entire litigation. Novelty and inventive step, the two general requirements administered in every patent system, were found in the applicant’s favour. The application failed anyway, on a provision that operates after those requirements are met. The Board also recorded, in discussing price, that during the period of exclusive marketing rights the appellant charged Rs 1,20,000 per month for a required dose of the drug, a figure it described as not disputed by the appellant, and it referred to the company’s patient assistance programme for free distribution to certain patients. The Supreme Court reproduced that passage in its judgment. Price therefore entered the case as a recorded observation of the tribunal below, not as a ground of the Supreme Court’s decision, and the distinction matters to any accurate account of what the judgment held.

Novartis filed special leave petitions in the Supreme Court on 11 August 2009, coming directly against the Board’s order rather than through the High Court. Two opponents, Natco Pharma and the Cancer Patients Aid Association, also filed petitions, directed at the findings the Board had recorded in Novartis’s favour. The Court, having heard the matter at length before an earlier bench in 2011, decided to determine the case itself rather than remit the parties to the High Court, while recording that the direct route was not to be treated as a precedent. It decided the case unbound by the findings of the authorities below, with all issues open before it. The route itself was exceptional in a second respect. Appeals from the Appellate Board ordinarily lay to the High Court, and the Supreme Court entertained the petitions directly, recording that the controversy had already travelled through every tier the system provides and had been argued before it at length, while making clear that the direct entertainment was not to be cited as a precedent for bypassing the High Court in patent matters. The petitions had in fact been argued at length before an earlier bench in August and September 2011, and the matter was then placed before the bench that decided it. The judgment that followed, delivered by Aftab Alam J for a bench also comprising Ranjana Prakash Desai J, is therefore best read as a first instance determination of the provision’s meaning by the final court, made on the full record.

The arguments the Court heard

The judgment’s structure becomes intelligible once the submissions are set out, because the Court accepted the opponents’ framework while deciding the case on narrower grounds than the opponents proposed, and rejected the appellant’s framework while accepting one of its factual premises. Four positions were before the Court.

For Novartis, leading counsel Gopal Subramanium advanced what may be called the architectural argument. Section 3, on this submission, does not add free standing conditions of patentability to the Act. Its clauses illustrate, for particular subject matter, what the definition of invention already excludes, and clause (d) in particular is directed at a mere discovery: a finding that adds nothing to the stock of working knowledge. The beta crystalline form, counsel submitted, was no mere discovery. It was arrived at by a two stage process of invention, first the selection of imatinib from the Zimmermann genus and its conversion into imatinib mesylate, itself an invention, and then the arrival at the beta crystalline form of the salt. On the facts, counsel submitted that the Zimmermann patent had disclosed the genus and the free base, not the mesylate salt as a known substance with known efficacy, and the United States authority in Hogan was pressed for the proposition that a generic disclosure does not make later species known. On efficacy, counsel submitted that the word bears its pharmaceutical meaning and is not confined to therapeutic effect in the narrow sense: a form that is absorbed more reliably, that delivers the active substance to the blood in greater measure, and that can be processed and stored dependably, works better as a medicine, and the affidavits established an improvement of about thirty per cent in bioavailability together with the three physical advantages. Finally, counsel pointed to the drafting asymmetry within the clause: the Explanation speaks of differences that are significant in properties with regard to efficacy, and a construction that confines efficacy to therapeutic effect read into the clause a limitation Parliament had not expressed in terms.

A second strand of the appellant’s case, with T. R. Andhyarujina also appearing for Novartis, addressed the relationship between Section 2 and Section 3 and the consequences of treating the clause as a fourth condition of patentability superimposed on the three the Act states generally. The submission’s force lay in its appeal to coherence: a statute that defines invention exhaustively and then excludes specified matters from the definition should not lightly be read as authorising the refusal of a patent for subject matter that answers the definition and escapes every specified exclusion properly construed.

For the respondents and the opponents, the submissions defended the clause as a deliberate and permissible legislative choice. India, it was argued, was entitled under the international framework to define the content of invention for itself, and Parliament had exercised that entitlement for a stated reason: the apprehension, recorded in the debates, that the return of product patents for medicines would be exploited by repetitive patenting of variants, extending monopolies without therapeutic gain. The clause was the chosen instrument against that apprehension, and its standard was efficacy because efficacy is what a medicine is for. On the facts, the opponents’ case was forensic. The application itself stated that all the indicated inhibitory and pharmacological effects are also found with the free base. The affidavits compared the wrong substances. The properties proved were the properties of the form as a material. Nothing in the record showed that a patient treated with the beta form fared better than a patient treated with the substance already known.

Anand Grover, for the Cancer Patients Aid Association, added the pharmacological submission described earlier in this article: that efficacy is a pharmacodynamic property, bioavailability a pharmacokinetic one, and that the two are conceptually distinct, so that evidence of the second kind can never answer a test framed in the first. Professor Shamnad Basheer, as intervenor and amicus curiae, proposed the middle position: that a construction of therapeutic efficacy should be capable of taking account of safety and of significantly reduced toxicity, since a form of a drug that patients can tolerate, or that harms them less, treats them better in a sense no clinician would dismiss. The Court’s disposition of these four positions is the judgment’s fingerprint. It adopted the respondents’ framework on the clause’s place in the Act. It resolved the factual issues on the opponents’ forensic ground, the identification of the known substance and the absence of any comparison against it. It accepted neither the pharmacodynamic submission nor the amicus’s safety submission, and it refused the appellant’s framework while leaving the inventive step finding of the Board untouched. Later practice cites the case for its holdings. Its method is visible only when the arguments are kept in view: the Court decided the least it needed to decide, on the record as it stood, and left the two most interesting questions in the case open for a record that would require them to be answered.

Two questions, not one: how Section 3(d) sits inside the Act

The judgment opens by separating questions that public discussion habitually merges. What is the true import of Section 3(d), the Court asked, and how does it interplay with clauses (j) and (ja) of Section 2(1)? Does the product qualify as a new product arrived at through an invention involving technical advance that makes it not obvious to a person skilled in the art? And if it does satisfy the definition of invention, can its patentability still be denied because Section 3(d) puts it outside the category of invention? The third question contains the article’s structural answer. The Court held that the distinction between invention and patentability lies at the heart of the Act as framed in 1970 and was reinforced by the 2005 amendments. Section 3(d), it said, clearly sets up a second tier of qualifying standards for chemical substances and pharmaceutical products, in order to leave the door open for true and genuine inventions but, at the same time, to check any attempt at repetitive patenting or extension of the patent term on spurious grounds.

Can a claim pass novelty and inventive step and still fail in India?

Yes. The Appellate Board found the beta crystalline form novel and not obvious, and the application failed anyway, because Section 3(d) imposes a further requirement on new forms of known substances after the general tests are met. Novelty and inventive step are necessary for such a claim. After Novartis, they are not sufficient.

One textual tension in the Act deserves notice here, because the two tier structure resolves it without mentioning it. The definition of inventive step in Section 2(1)(ja) is generous to commercial advantages: a feature may qualify if it involves technical advance as compared to existing knowledge, or if it has economic significance, or both. Read alone, that definition might appear to admit the manufacturing and handling virtues of a new form as part of the inventive inquiry. Section 3(d) forecloses any such route for new forms of known substances. Whatever economic significance a polymorph’s flow or stability may carry within the inventive step analysis, the clause superimposes a test on which those virtues count for nothing, because they are not efficacy. The two provisions therefore operate in sequence and, within their sequence, in different currencies: the first asks whether the advance was real and non-obvious, in technical or economic terms; the second asks the narrower question what the advance did for the substance’s known working. A claim must be solvent in both currencies. The judgment’s second tier metaphor is the simplest accurate description of that arrangement, and it explains why the parties’ dispute about whether the clause is a definition or an exclusion mattered less than the clarity the metaphor supplied.

The Court also identified a second way of reading the relationship, and its candour about the alternative is part of the judgment’s value. Viewed within Section 3 as a whole, the clause states a ground of patentability, distinct from the definition of invention. Viewed in isolation, the clause might instead be read as an extension of the definition itself, setting a higher invention threshold for medicines, drugs and other chemical substances than for other subject matter. The Court did not need to choose between the two readings to decide the case, because on either reading the same product had to pass the same additional test. It stated the outcome in the alternative form in its concluding analysis: in the case of chemicals and especially pharmaceuticals, if the product claimed is a new form of a known substance with known efficacy, then the subject product must pass, in addition to clauses (j) and (ja) of Section 2(1), the test of enhanced efficacy as provided in Section 3(d) read with its Explanation.

Three gates for a pharmaceutical claim

The table states the sequence as the judgment leaves it. It is the findable artifact of this article because it makes visible the feature of Indian law that applicants from other jurisdictions most often miss: the third gate exists, it is administered after the first two, and it has its own evidence.

Gate	Question the gate asks	What must be established	What satisfies it	Illustration
Gate one: novelty	Is the claimed subject matter new?	The claim is not anticipated by publication or prior use; it does not form part of the state of the art	A specification that distinguishes the claim from the prior art on the face of the documents	The Appellate Board held the beta crystalline form novel: no earlier document disclosed that polymorph as such
Gate two: inventive step	Is the advance not obvious?	A feature involving technical advance over existing knowledge, or economic significance, or both, that makes the claim not obvious to a person skilled in the art, under Section 2(1)(ja)	Evidence and argument directed at what the skilled person would have done with the prior art, not merely at what the art contains	The Appellate Board held the selection and preparation of the beta form not obvious, reversing the Assistant Controller
Gate three: Section 3(d)	Is a new form of a known substance shown to work better as a treatment?	Where the claim is a new form of a known substance with known efficacy, enhancement of that known efficacy, which for a medicine means enhanced therapeutic efficacy	Research data in or supporting the specification, comparing the claimed form against the known substance it derives from, directed at therapeutic effect	The beta crystalline form failed: the comparison offered was against the free base and the alpha form, and the properties proved were flow, stability and moisture behaviour, not therapeutic effect

The sequence in the table is analytical rather than strictly chronological, and the qualification matters in practice. Novelty and inventive step are examined on the prior art; the third gate may be raised at the same examination, in opposition before or after grant, or in revocation and infringement proceedings years later, and the evidence relevant to it, the specification’s own statements and the comparative data, is fixed at filing in a way the prior art inquiry is not. An applicant therefore cannot sequence its own preparation in the order of the table. The third gate’s record must be built first, when the application is drafted, although the gate itself may be encountered last.

Read down the rows and the litigation explains itself. The applicant won the second row outright before the Board and was not required to win it again in the Supreme Court, which expressly left the inventive step question aside once the known substance finding was made. The case was decided in the third row, on two failures that compounded: the comparison was run against the wrong predecessor, and the properties proved were of the wrong kind.

Two thresholds in one clause: enhancement, and significance

A close reading of the clause discloses two verbal formulas, and the judgment uses both without treating them as interchangeable. The substantive limb requires that the new form result in the enhancement of the known efficacy of the known substance. The Explanation requires that a derivative differ significantly in properties with regard to efficacy before it escapes being deemed the same substance. The formulas do different work. The Explanation’s formula is the prior one in the analysis: it fixes identity. Unless the derivative differs significantly in the stated respect, the law treats it as the known substance itself, and a claim to the known substance is a claim to what is already known. The substantive limb’s formula is the operative test once the claim is analysed as a claim to a new form: enhancement of the known efficacy must result from the form. Read together, they impose a double discipline on the applicant’s evidence. The difference asserted must be a difference in properties with regard to efficacy, which excludes the physical dossier. And the difference must be significant, and result in enhancement, which excludes the marginal dossier: a difference in therapeutic effect that is real but trivial does not carry a claim over either formula.

The burden architecture follows from the deeming structure. In an ordinary patentability dispute the office or the opponent must establish that a claim is anticipated or obvious. Under Section 3(d) the statute first deems the derivative to be the same substance, and the applicant who asserts a significant efficacy-linked difference must establish it. That is why the judgment’s evidentiary holdings are expressed as failures of proof rather than as findings against the applicant: the Court did not hold that the beta form lacked enhanced therapeutic efficacy as a matter of chemistry. It held that the material placed before it did not establish enhancement against the relevant baseline, and that the burden of placing such material lay where the clause puts it, on the party claiming the patent. Applicants who treat the clause as an objection to be argued away, rather than as a deeming provision to be displaced by data, repeat the error the case records.

The Court did not find it necessary to decide whether the two formulas state one standard or two, and commentators divide on the question: whether “differ significantly in properties with regard to efficacy” adds a quantitative floor to the qualitative requirement of enhancement, or merely restates it from the direction of identity. The safer reading of the judgment is that the Court treated the Explanation as an aid to the substantive limb, using it to identify what does not count as efficacy, rather than as a separate hurdle with its own measure of significance. The point is flagged here because it is live in examination practice, where examiners invoke both formulas, and because a future case with a genuine but modest therapeutic difference will require it to be answered.

The clause among its neighbours in Section 3

Section 3 is a miscellany, and the clause is better understood when its neighbours are in view, because the Act distributes the policing of pharmaceutical claims among several of them and a filing that escapes one may be caught by another. Clause (c) excludes the mere discovery of a scientific principle or the formulation of an abstract theory, and the discovery of any living thing or non-living substance occurring in nature. Clause (d) presupposes that family resemblance: the new form is treated as something found in or derived from what exists, and the clause’s question is whether what was found works better. Clause (e), examined in the answers below, excludes mere admixtures whose properties are only the aggregation of their components. Clause (i) excludes processes for the medicinal, surgical, curative, prophylactic, diagnostic, therapeutic or other treatment of human beings or animals. A pharmaceutical applicant in India therefore faces a section, not a clause: the active ingredient’s new form under (d), the formulation’s combination under (e), the treatment regimen under (i), and the naturally occurring substance under (c), each with its own question and its own evidence.

The ordering of these questions was the point the Delhi High Court addressed in the Erlotinib litigation when it described the Section 3 exclusions as threshold conditions of patent eligibility rather than exceptions to the definition of invention. The distinction is practical. If the clauses were exceptions, an applicant would first establish an invention under Section 2 and the burden would then shift to an opponent to bring the claim within an exclusion, narrowly construed. If they are threshold conditions, the applicant must show at the outset that the claim is the kind of subject matter the Act entertains at all, and the exclusions are construed as part of the definition of the field. The Supreme Court in Novartis had expressly left the choice between the alternative readings open, deciding the case on the footing that the same test applied either way. The Delhi High Court’s threshold reading has since ordered examination and pleading practice, and it aligns with the deeming structure of clause (d) itself: a provision that begins by deeming a derivative to be the same substance as its parent is naturally administered as a threshold, since until the deeming is displaced there is, in the eye of the statute, no new subject matter to examine.

The first holding: identifying the known substance

Everything in the case turned on a prior question that reads as housekeeping and is in fact the decision: of which known substance was the beta crystalline form a new form? Novartis’s case, as presented, offered two routes. In the application and the affidavits, the beta form was presented as derived from imatinib in free base form, the substance of the Zimmermann patent. In argument before the Court, senior counsel presented the product as two stages removed from the free base: a first invention selecting imatinib and converting it to its methanesulfonic acid salt, imatinib mesylate, and a second invention arriving at the salt’s beta crystalline form. The two presentations were not consistent with each other, and the Court noticed. The application, it observed, contains references only to the free base and to the alpha form; the impression it gives is that the beta form is derived directly from imatinib.

The Court then read the Zimmermann patent for itself and made the finding on which the rest of the judgment rests. Imatinib mesylate, it held, is a known substance from the Zimmermann patent itself. The earlier specification described the compounds together with their acid addition salts, explained that a reference to a free compound includes its salts where appropriate, and described the pharmacological properties of what it disclosed, properties also carried in an article in the journal Cancer Research. The mesylate salt was therefore not a substance awaiting invention in 1997. It was within the disclosure of 1993 and 1994, and its efficacy, its capacity to inhibit the kinases implicated in the disease and to act against tumours, was known with it. Two consequences followed. Imatinib mesylate did not itself qualify as an invention under Section 2(1)(j) and (ja), since a substance disclosed in the prior patent cannot be a new product. And for the purposes of Section 3(d), the known substance against which the beta form fell to be compared was imatinib mesylate in its non-crystalline state, the substance immediately preceding the claimed form.

Novartis resisted the finding with an argument drawn from United States authority, the decision in Hogan concerning polypropylene, under which a patent claiming a genus may cover species neither known nor enabled when it was filed. The Court declined to import it. The finding that imatinib mesylate was known rested, it said, on an objective consideration of the material, not on the applicant’s conduct; and it added a passage that reveals its anxiety about genus claiming generally, observing that it did not wish patent law in this country to develop on lines where the scope of a patent is determined not on the intrinsic worth of the invention but by the artful drafting of its claims by skilful lawyers. The genus and species problem, of which the Hogan argument is one expression, had been placed before the Court and was answered on the facts: whatever the position may be for a genuinely undisclosed species, the mesylate salt was disclosed, as a salt, with its properties, and that ended the inquiry.

The Court supported the identification from the Act’s own usage of the word known. In the revocation provisions, Section 64(1)(e) and (f) speak of a claimed invention being publicly known or publicly used, and of subject matter known from a document, and the Court examined its earlier decision in Monsanto Company v Coramandal Indag Products on the construction of that language. The point of the excursion was consistency: a substance does not cease to be known, for the purposes of an Act that repeatedly uses the concept, because the document that discloses it describes it within a family, in company with its salts, rather than by a separate name and a separate example. Knowledge in patent law attaches to what the disclosure makes available to the skilled reader, and the Zimmermann disclosure made imatinib mesylate available, with the properties that made it worth developing. The Court added that its conclusion was in no way based on the appellant’s conduct: the identification was objective, a reading of documents, not a penalty for the way the application had been presented, although the application’s own references to the free base and the alpha form made the two stage presentation difficult to sustain on the file.

A distinction embedded in the identification deserves emphasis, because conflating it with anticipation produces most of the confusion in commentary on the case. To say that imatinib mesylate was a known substance for Section 3(d) is not to say that the claimed beta crystalline form was anticipated. Anticipation is a doctrine about claims: a claim is anticipated when a single prior document discloses every feature of what is claimed, so that the claim adds nothing. The Appellate Board held that no document disclosed the beta form as such, and on that footing found novelty, a finding the Supreme Court did not disturb. Identity under Section 3(d) is a doctrine about substances: the Explanation deems a derivative to be the same substance as a known substance unless a significant efficacy-linked difference is shown, and the Zimmermann disclosure established which substance was known. The same document therefore performed two different operations in the case. It failed to anticipate the claim, because it did not disclose the claimed form. It succeeded in identifying the known substance, because it disclosed the salt from which the claimed form derived, together with the salt’s properties. A prior patent can thus lose on novelty and win the case on Section 3(d), and any analysis that treats the two holdings as inconsistent has mistaken the second operation for the first.

The comparison discipline that follows from the identification is the holding’s practical core, and it is the element of the judgment most often missed in accounts of the case. Once the known substance is imatinib mesylate, an applicant must show the enhanced efficacy of the new form over imatinib mesylate. Novartis had shown no such thing. Its application compared the beta form’s flow, stability and hygroscopicity against the alpha crystalline form. Its affidavits compared solubility and bioavailability against the free base. There was, the Court recorded, no material in the application or the affidavits making any comparison of efficacy, or even of solubility, between the beta crystalline form and imatinib mesylate in non-crystalline form. The Court further observed that the higher solubility attributed to the beta form might in truth be a property of imatinib mesylate itself, since salts normally have much better solubility than compounds in free base form, so that the only properties safely attributable to the beta form as such were the three physical attributes: better flow, better thermodynamic stability and lower hygroscopicity. A case built on comparisons against the wrong substances had left the legally relevant comparison unmade, and no advocacy could repair the record. That is why the identification of the known substance decided the case before the meaning of efficacy was reached: it fixed the baseline, and against that baseline there was no evidence at all.

The second holding: efficacy means therapeutic efficacy

The Court approached the undefined word through its ordinary meaning and the function of the product under consideration. Efficacy means the ability to produce a desired or intended result. The test of efficacy in Section 3(d) therefore depends on the result the product is intended to produce, which is to say on its function, utility or purpose. For a medicine that claims to cure a disease, the test of efficacy can only be therapeutic efficacy. The Court reinforced the reading from the provision’s genesis and from its internal structure. Given the circumstances in which the clause was amended to make it even more constrictive than before, the therapeutic efficacy of a medicine must be judged strictly and narrowly. The substantive limb demands enhancement of the known efficacy, and the Explanation demands that a derivative differ significantly in properties with regard to efficacy. What is evident, the Court said, is that not all advantageous or beneficial properties are relevant, but only such properties that directly relate to efficacy, which in the case of a medicine is its therapeutic efficacy.

One refinement of the holding is routinely overlooked and repays attention, because it fixes the content of the comparison from the other side. The clause requires enhancement of the known efficacy, and the word known does double work. It identifies the substance, and it identifies the efficacy to be enhanced: the efficacy that was known for that substance. For imatinib mesylate the known efficacy was its anti-tumour, kinase inhibiting working, the property the Zimmermann specification and the Cancer Research literature had made known. Enhancement therefore had to be shown in that working, or in the therapeutic outcome it produces, and not in any property the form happened to improve. An applicant who proves that a form is a better material has not begun to answer the question, however rigorously the proof is assembled, because the question is set by the efficacy the substance was already known to have. This is also why the Court could treat the case as concluded by the identification of the known substance together with the record: once the known efficacy was fixed as the anti-tumour working of imatinib mesylate, the entire evidentiary record, directed at flow, stability, moisture and absorption, was pointed at targets the statute does not set.

The Explanation received a construction of its own that applicants ignore at their cost. Each form listed in it has properties inherent to that form: solubility to a salt, hygroscopicity to a polymorph. A change of form that produces only the properties inherent in the new form does not, without more, amount to enhancement of efficacy, because the Explanation is meant to indicate what is not to be considered as therapeutic efficacy. In other words, the statute anticipates the standard dossier of polymorph advantages and disqualifies it in advance. A polymorph that flows better, keeps better and absorbs less moisture has demonstrated exactly the properties a polymorph is expected to have, and those properties, the Court held in the plainest sentence of the analysis, cannot even be taken into account for the purpose of the test of Section 3(d), since these properties have nothing to do with therapeutic efficacy. The properties may make a substance easier to process, easier to store and cheaper to handle, and those are real benefits in manufacture. The judgment does not deny them. It classifies them. They belong to the substance as a material, not to the medicine as a treatment, and the clause asks only about the medicine as a treatment.

The classification the holding imposes can be tested against the standard repertoire of follow-on claims, and doing so shows how much of the repertoire it displaces. A new salt chosen because it crystallises in a form that can be milled, blended and compressed reliably shows manufacturing virtues, not therapeutic ones. A polymorph that survives hot and humid storage without converting shows a stability virtue, valuable in a country where medicines travel far and cold chains are uneven, and still, on the holding, not a therapeutic one, because the advantage attaches to the substance as a material and to the supply chain rather than to the treatment of the patient. A particle size reduction that makes a poorly soluble drug dissolve faster stands nearer the line, because dissolution feeds absorption and absorption feeds effect, and the judgment’s treatment of bioavailability governs it: the connection must be proved through to therapeutic benefit, not presumed from the dissolution curve. A formulation that reduces a toxic metabolite, or a form that patients can actually complete a course of treatment on, stands on the other side of the line in substance, although, as the next paragraph records, the Court left the status of safety and tolerability undecided, so that claims in that territory must be advanced as argument rather than as settled law. The map the judgment draws is therefore not a map of worthy and unworthy research. It is a map of kinds of proof: proof about the material on one side, proof about the patient on the other, and only the second kind answers the clause.

Two refinements in the argument were recorded and deliberately left open, and they mark the live frontier of the doctrine. Counsel for the Cancer Patients Aid Association, Anand Grover, submitted that efficacy in pharmacology is a pharmacodynamic property, the capacity of a drug to produce its effect at its target, and that bioavailability, a pharmacokinetic property concerned with the fraction of a dose reaching its site of action, is a different thing altogether, so that no showing about bioavailability could ever answer the clause. Professor Shamnad Basheer, appearing as intervenor and amicus curiae, accepted that properties such as processability and flow cannot qualify, but argued the less rigid position that safety, or significantly reduced toxicity, should also be taken into consideration in judging enhanced therapeutic efficacy. The Court recorded both submissions with respect and made no pronouncement on either, holding that the case could be finally and effectively decided without them. The status of safety and toxicity under the clause is therefore not settled by Novartis, and any account that reports the judgment as having decided it misreports the case.

The third holding: bioavailability must be proved, not assumed

Why did the claim of thirty per cent higher bioavailability not save the application?

The Court held that increased bioavailability does not by itself establish enhanced therapeutic efficacy, and that any such link must be claimed and proved by research data. Novartis relied on a rat study showing roughly thirty per cent greater absorption, but it compared the beta form against the free base and produced nothing connecting the increase to a therapeutic benefit.

The holding’s operative sentence repays parsing, because it imposes two requirements and applicants have failed on each separately. The enhancement must be specifically claimed: the specification must itself assert, as part of the invention presented, that the form enhances therapeutic efficacy, so that the assertion is part of what is examined and part of what the public is told the patent is for. And it must be established by research data: the assertion, however specifically made, does not discharge itself. Novartis failed the second requirement outright, since the link between bioavailability and therapeutic benefit was asserted in argument and nowhere established. Later applicants have failed the first, filing specifications that describe a form and its physical virtues in detail while leaving the therapeutic case to be assembled in prosecution correspondence, a sequence the sentence forecloses in principle: a benefit that is neither claimed in the document nor established by data is, for the purposes of the clause, not part of the application at all. The sentence thus functions as a rule of pleading as much as a rule of evidence, and it explains the transformation of drafting practice described later in this article, in which the efficacy narrative has migrated from the reply to the examination report into the specification itself.

The evidentiary trail matters and the judgment reconstructs it carefully. After the law changed while the application waited in the mailbox, Novartis filed four expert affidavits. Two addressed experiments conducted for the opponents. The other two, by Paul William Manley and Giorgio Pietro Massimini, were filed to meet Section 3(d). Massimini’s affidavit explained candidly that the provision was unique to India, that no study of the free base’s efficacy had previously been carried out because no occasion had arisen, and that a study had been commissioned once the requirement became known. The study, conducted in rats, produced an area under the curve comparison from which about thirty per cent improvement in bioavailability for the beta crystalline form over the free base was derived. Manley’s affidavit addressed solubility, recording that the free base is very slightly soluble in water while the mesylate salt is very soluble, and repeated the three physical advantages over the alpha form.

Working against the affidavits was the applicant’s own specification, and the Court’s use of it shows how completely the case was lost on the documents. The application of 1998, describing the state of knowledge it built on, stated that all the indicated inhibitory and pharmacological effects are also found with the free base of the active substance. In 1998 the sentence was unremarkable: it situated the claimed form within a known pharmacology and claimed for the form only its physical advantages. After 2005 the sentence read as a concession that the form’s therapeutic working was the known working, unchanged. An applicant cannot easily retreat from its own specification, which is the document the statute requires to describe the invention fully and particularly, and the Court treated the statement as part of the objective record rather than as a slip to be explained away. The lesson generalises beyond the case. In a Section 3(d) prosecution, the specification is not a neutral container for the claims. Its background section, its statements of advantage and its choice of comparators are evidence, and they are the first evidence the examiner and the court will read.

Measured against the holdings already made, the affidavits failed twice over. They compared the claimed form against imatinib free base, which was not the known substance the Court identified, and against the alpha crystalline form, which was not the known substance either. And the one property they quantified, bioavailability, was a property whose connection to therapeutic efficacy the judgment refused to presume. The Court’s sentence is the operative rule for every later applicant: just increased bioavailability alone may not necessarily lead to an enhancement of therapeutic efficacy, and whether or not an increase in bioavailability leads to such an enhancement in any given case must be specifically claimed and established by research data. There was, the Court found, nothing at all on this score apart from the submissions of counsel, and no material indicating that the beta form would produce an enhanced or superior therapeutic efficacy on a molecular basis over what imatinib free base achieved in the animal model relied on. The failure was therefore not a failure of chemistry. The beta form may well be better absorbed. The failure was a failure of proof, on a record assembled under time pressure, against a baseline the applicant had misidentified, in answer to a question the applicant had not been required to address when it filed. The judgment is unsentimental about the consequence, and the consequence is the lesson the profession took from the case: in India, the efficacy case for a new form must be built when the application is drafted, against the substance the statute will identify, and with data that speaks to therapeutic effect.

A paragraph on the package, and what it shows about the claim

One further passage of the judgment deserves separate statement because it is rarely quoted and it explains the Court’s tone. The drug marketed in the United States in 2001, and in India under exclusive marketing rights from November 2003, was imatinib mesylate. The package described the product as imatinib mesylate tablets of 100 milligrams, each film coated tablet containing 100 milligrams of imatinib as mesylate. There was no reference on the package to imatinib mesylate in beta crystalline form. From this the Court drew an inference about the claim itself: if what was sold throughout was the salt, a claim to the salt’s crystalline form, pursued after the salt could no longer be claimed, would only appear as an attempt to obtain a patent for imatinib mesylate, which would otherwise not be permissible in this country. The passage is not a holding and the decision does not rest on it. It is included here because it shows how the Court understood the commercial anatomy of the application: the product in the pharmacy, the substance in the earlier patent, and the form in the later claim were, in its analysis, three presentations of one medicine, and the law it was construing existed to police the distance claimed between them. The passage also performs quiet corroborative work for the central holding. The finding that imatinib mesylate was the known substance, and the beta form merely its new form, was a finding about documents: the Zimmermann disclosure on one side and the application on the other. The package showed the same relationship in the marketplace: what physicians prescribed, what patients consumed and what the label described was the salt, in tablets whose contents were stated in terms of the salt, and the crystalline state that the patent claim singled out appeared nowhere in the transaction between the company and the patient. A court construing a provision directed at the substance of claims rather than their form could hardly have asked for a plainer illustration of the gap the clause exists to close.

What the Court did not decide

A judgment’s authority is fixed as much by its refusals as by its holdings, and Novartis contains four refusals that later argument often overlooks. The first concerns incremental innovation. The Court stated expressly that its decision is not to say that Section 3(d) bars patent protection for all incremental inventions of chemical and pharmaceutical substances, and that it would be a grave mistake to read the judgment to mean that the clause was amended with the intent to undo the fundamental change brought about by the deletion of Section 5. Product patents for pharmaceuticals exist in India; the clause qualifies them; it does not repeal them. Genuine incremental invention, demonstrated by the evidence the clause demands, remains patentable. Any summary of the case that drops this paragraph has changed the judgment’s meaning.

The second refusal concerns TRIPS. The Supreme Court recited the history of the agreement and the Doha Declaration at length, but as context for the legislative choices Parliament made, recording that it had referred to the agreement not to comment on its fairness or to examine the wisdom of subscribing to it. The compliance question had been placed before the Madras High Court, which declined it as a matter for the World Trade Organization’s dispute settlement system, and Novartis had not appealed that decision. The Supreme Court accordingly made no finding that the provision complies with TRIPS and no finding that it does not. The question stands where the treaty system leaves it: available to any member state that chooses to raise it in Geneva, and never raised.

The third refusal concerns the internal debate about efficacy recorded above. The Court left undecided both the submission that efficacy is strictly a pharmacodynamic property to which bioavailability is irrelevant in principle, and the submission that improved safety or reduced toxicity can constitute enhanced therapeutic efficacy. The doctrine’s centre is fixed; its edges are not. The safety question is the edge with the largest practical territory, because a substantial share of genuine pharmaceutical improvement consists in making a known treatment tolerable: reducing the toxicity that forces patients to abandon a course, widening the margin between the effective dose and the harmful one, or removing an interaction that excludes a class of patients. Applicants in that territory argue, following the line the amicus proposed, that a form a patient can complete treatment on is therapeutically more effective than one the patient cannot, and that efficacy judged by outcomes rather than by mechanism should receive the argument. Opponents answer that the judgment confines efficacy to the treatment’s effect on the disease, and that tolerability is a property of the regimen rather than of the substance’s known efficacy. No appellate decision has settled the exchange, and the question waits for a case in which safety is the only difference and the difference is large, the case Novartis, with its three physical properties, never was. The fourth refusal is quieter. The Court assumed, for the sake of argument, that the beta form might be accepted as new, and expressly declined to decide whether it involved an inventive step. The Appellate Board’s finding of non-obviousness was therefore neither endorsed nor displaced, and the case must not be cited as a Supreme Court determination that the claimed form was, or was not, obvious.

The formal outcome completes the picture. The Court dismissed Novartis’s appeals with costs, and allowed the appeals filed by Natco Pharma and the Cancer Patients Aid Association. The application that had entered the mailbox in 1998, and had supported exclusive marketing rights and infringement enforcement for a period in between, ended without a product patent, fifteen years after it was filed and eight years after the provision that decided it had taken its amended shape.

The two cases about the provision, stated at full strength

The judgment decided the case. It did not end the argument about the clause, which continues between two positions that this article states in their strongest form, attributed to those who make them, and does not join.

The originator case begins with the character of pharmaceutical research. Valuable advances in medicines are frequently incremental: a salt, a polymorph, a formulation or a dosing form that is more stable, safer in handling, easier to manufacture at scale, more consistent in absorption, or suitable for patients who cannot use an earlier form. Each such advance costs research money, and each is exposed to immediate copying once published. A rule that withholds patent protection unless the advance improves therapeutic effect, proven by research data at the filing stage, removes the incentive to make and disclose such advances at all, or drives them into trade secrecy, and patients lose improvements that a looser rule would have delivered. The case continues with a complaint about the standard itself. Parliament used the words efficacy, enhancement and significantly without defining any of them, and the content the word received, therapeutic efficacy judged strictly and narrowly, was supplied by a court eight years after the provision took effect and fifteen years after the application in the case was filed. A standard whose meaning is unknowable when research decisions are taken cannot order investment, and the retroactive quality of the exercise, where an application drafted under one regime is measured against a test articulated under another, is presented as a fairness objection independent of the medicine involved. Novartis’s own evidence made the related point that the provision has no counterpart elsewhere: its expert recorded that the requirement was unique to India and that there was no analogous provision in the patent law of any other country. The originator case ends in international law. Article 27.1 of TRIPS requires that patents be available for any inventions, whether products or processes, in all fields of technology, provided they are new, involve an inventive step and are capable of industrial application, and that patent rights be enjoyable without discrimination as to the field of technology. A fourth criterion, applied in practice to one industry, however it is labelled, is said to breach both halves of that sentence. The United States Trade Representative has kept India on the Priority Watch List in its Special 301 reports and has recorded industry concern about the provision and about the criteria applied to pharmaceutical patents, and trade associations of research based companies continue to press the point in that forum.

The access case answers each element. On the text of the treaty, TRIPS obliges members to grant patents for inventions but nowhere defines invention, novelty or inventive step; the content of those concepts is left to national law, and a provision defining what counts as an invention for one class of subject matter is an exercise of that reserved competence, not an addition to the treaty’s criteria. Discrimination is answered from the clause’s own words: it is directed to chemical substances as a class, a drafting choice the Supreme Court recorded, and a rule about discoveries as against inventions applies a distinction the Act draws throughout Section 3, not a distinction invented for one company’s product. The Doha Declaration on the TRIPS Agreement and Public Health, adopted by ministers in November 2001, is invoked for the proposition that the agreement does not and should not be interpreted and implemented in a manner that prevents members from protecting public health and promoting access to medicines for all, and the Indian provision is presented as a flexibility of exactly the kind the declaration confirmed. On the merits, the access case rests on the practice the clause was drafted against. Evergreening, the strategy of obtaining successive patents over salts, polymorphs, formulations and doses of one known drug, each timed to take effect as an earlier patent expires, can extend a monopoly over a single medicine far beyond the twenty year term the treaty contemplates, without any corresponding advance in treatment. A filter that admits only variants shown to treat better, and admits those freely, is characterised as the calibrated response: it does not touch the original product patent, it does not shorten any term, and it asks of a second patent only that the second invention be real. The judgment’s own reservation, that genuine incremental inventions remain patentable and that reading the case otherwise would be a grave mistake, is quoted back as the answer to the charge of hostility to incremental research. Médecins Sans Frontières, which supplies medicines in the developing world, welcomed the 2013 judgment publicly as a protection for access to affordable medicines, and patient organisations make the same argument from the price record in the case itself.

One legal strand of the debate deserves separate statement because it recurs wherever the provision is discussed internationally. The Doha Declaration, in its fifth paragraph, enumerates particular flexibilities: the freedom to grant compulsory licences and to determine the grounds for them, the right to determine what constitutes a national emergency, and freedom on exhaustion of rights. The originator side observes that a patentability filter of the Indian kind appears nowhere in that enumeration, and argues that the flexibilities TRIPS permits are those the agreement and the declaration specify. The access side replies that the enumeration addresses flexibilities within a functioning patent system, while the definition of what counts as an invention at all is a prior question the agreement reserves to members by silence: Article 27 obliges members to make patents available for inventions without defining the term, and a state that defines the term with an efficacy condition for derivatives has not breached a text that nowhere forbids the definition. The Supreme Court’s refusal to pronounce on compliance leaves both arguments intact and untested, and they are rehearsed, in substantially these forms, in every multilateral discussion of the provision.

Beneath the legal formulations, the two cases disagree about a factual prediction, and the disagreement is worth making explicit because no court can settle it. The originator case predicts behaviour: that a demanding efficacy threshold, administered with the benefit of hindsight over applications drafted years earlier, teaches research based companies to discount India in filing and launch decisions, and teaches their scientists that improvements in delivery, stability and tolerability, which clinical practice values, will go unrewarded in one of the world’s largest medicines markets. The access case predicts the opposite behaviour under the opposite rule: that without an efficacy filter, the economics of follow-on patenting dominate, because a variant patent is cheaper to obtain than a new medicine is to discover, litigation over a thicket of variant patents is cheaper for an incumbent to fund than for a generic entrant to clear, and each year of deferred competition over a high volume medicine transfers sums that dwarf the research cost of the variant itself. Both predictions are about incentives under uncertainty, both are advanced by serious people with evidence of instances, and neither has been tested against a counterfactual India that chose the other rule. The clause represents Parliament’s resolution of that predictive dispute in 2005, and the judgment represents a court’s refusal to reopen the resolution in the guise of construction. That division of labour, legislature choosing the policy and court construing the instrument, is itself one of the judgment’s holdings in substance, visible in its repeated return to the debates and to the declared purpose of the amendment.

Two points of agreement are worth isolating because they define what the dispute is actually about. Both sides accept that the compliance question has never been determined in the only forum whose determination binds internationally: no complaint against the provision has been filed in the World Trade Organization’s dispute settlement system, and its compliance therefore remains undetermined there. And both sides accept that the clause leaves genuine pharmaceutical invention untouched: a new molecule, a new mechanism, a variant with demonstrated therapeutic advantage, all remain patentable. The contested ground is narrower than the rhetoric on either side suggests. It is the band of follow-on filings between the plainly genuine and the plainly strategic, and the dispute is about whether an efficacy filter, administered by examiners and courts on evidence, sorts that band more accurately than the ordinary tests of novelty and obviousness administered alone. Sticker prices, treaty texts and research budgets are the vocabulary of that dispute. Its substance is institutional: which test, applied by whom, on what evidence, makes the fewest errors of each kind.

The provision at work since the judgment

The judgment’s most immediate effect was administrative. In October 2014 the Office of the Controller General of Patents, Designs and Trade Marks issued guidelines for the examination of patent applications in the field of pharmaceuticals, which instruct examiners that an incremental invention based on a known substance with established medicinal activity will be deemed to fall foul of patentability where it fails to demonstrate significantly improved therapeutic efficacy as compared with the known compound, the language tracking the judgment’s construction of the Explanation. Examination practice accordingly treats the efficacy case as part of the application rather than an argument to be developed in correspondence: comparative data, directed at therapeutic effect and against the known substance, is expected in or with the specification, and examiners raise the clause as a standard objection against derivative claims.

The guidelines also work the clause through illustrations, applying the enhanced efficacy inquiry to the recurring categories of pharmaceutical filing: salts and polymorphs of known compounds, combinations, dosage forms and new uses, and directing examiners to ask, in each category, what therapeutic advantage over the known substance the specification demonstrates. The second working forum is the pre-grant opposition. Because any person may oppose an application before grant, and because the clause turns on documents and data rather than on market conduct, it is the ground best suited to that stage: an opponent with the prior patent, the applicant’s specification and the pharmacological literature can plead the entire objection without discovery or inspection. The Glivec prosecution, in which five opponents forced the question to a full hearing before the Assistant Controller, supplied the template, and the clause’s subsequent history has been written as much in opposition proceedings before Controllers as in court. Post-grant, the same objection travels into revocation petitions and into invalidity defences in infringement suits, where the patentee who obtained a grant without a contested efficacy record may find the question reopened on a fuller record than the examiner ever saw.

Beyond the courtroom, the judgment has a quieter administrative afterlife that determines its real reach. Patent examination in India is conducted by examiners working to guidelines and office practice, and the clause’s daily meaning is the meaning given to it in first examination reports across the four patent offices. There, the Novartis holdings have hardened into a checklist discipline: identify the known substance from the closest disclosure; ask whether the claim is to a form within the Explanation’s list; require the efficacy comparison against that substance; and treat physical parameter data, however extensive, as non-responsive unless a therapeutic bridge is built from it. Applicants respond in kind, and the prosecution of a pharmaceutical derivative application in India has become a distinctive genre of file, in which the specification’s examples are comparative, the comparators are chosen with the Zimmermann discipline in mind, and the efficacy narrative is drafted before the claims. A caution about the guidelines’ status completes the administrative picture. Examination guidelines are instructions to examiners, not legislation, and they bind neither applicants nor courts. Their Section 3(d) passages are an office synthesis of the judgment, useful because they show how the holdings have been operationalised, but where a guideline’s formulation and the judgment’s language diverge, it is the judgment that governs, and applicants and opponents argue from (2013) 6 SCC 1 rather than from the manual. The hierarchy matters in both directions. An examiner who refuses a claim by applying a guideline example more rigidly than the judgment warrants can be corrected on appeal by the judgment’s own reservations, in particular its refusal to bar incremental invention as such and its silence on safety and toxicity. Equally, an applicant cannot convert a guideline’s silence into a concession. The durable law of the clause remains the small set of holdings the Court actually made, administered through a large administrative apparatus whose instructions are evidence of practice and nothing more.

The clause has also been applied beyond medicines in form, since by its terms it governs chemical substances generally, although its litigation history is overwhelmingly pharmaceutical. One illustration from the record shows the test operating in the same register as the leading case. In the long running Erlotinib litigation between Roche and Cipla, the Delhi proceedings recorded that the Controller had refused Roche’s application for the B polymorph of Erlotinib hydrochloride on the ground that its increased thermodynamic stability did not meet the enhanced therapeutic efficacy standard of Section 3(d). The parallel with the Glivec claim is exact: a polymorph, a physical advantage in stability, and a refusal because the advantage proved was not a therapeutic one. The same litigation produced the clearest later statement of the clause’s place in the Act, when a Division Bench of the Delhi High Court held that the exclusions in Section 3 are not in the nature of exceptions to the definition of invention but are threshold conditions of patent eligibility, to be applied before the invention inquiry proceeds. That sequencing question, which the Supreme Court had deliberately left in the alternative, was answered in the Division Bench’s reasoning in favour of the threshold reading, and examiners administer the clause accordingly, at the threshold rather than as a refinement of obviousness.

The forum for disputes has changed since the judgment. The Intellectual Property Appellate Board, whose 2009 order was the decision under appeal in Novartis, was abolished with effect from 4 April 2021 by the Tribunals Reforms (Rationalisation and Conditions of Service) Ordinance, later enacted as the Tribunals Reforms Act, 2021, and appeals and revocation proceedings in patent matters lie to the High Courts. The Delhi High Court, which received the largest share of the transferred patent docket, constituted an Intellectual Property Division for the purpose and notified dedicated rules, the Delhi High Court Intellectual Property Rights Division Rules, 2022, on 24 February 2022, together with rules governing patent suits that expect parties to plead validity challenges with specificity, claim by claim and document by document. A Section 3(d) challenge in that forum is therefore no longer a general assertion that a patent covers an evergreened variant: it must be built, like the objection in Novartis itself, on the identity of the known substance and the absence of efficacy evidence, pleaded against identified claims. The wider map of forums, registers and rights across the intellectual property statutes is set out in the complete guide to intellectual property law in India, which this article does not duplicate. What the change means for the clause is continuity of substance through a change of forum: the standard set in 2013 is administered in examination and contested in the High Courts, and no appellate decision has displaced its central holdings, that efficacy in a medicine means therapeutic efficacy, that the comparison runs against the known substance immediately preceding the claimed form, and that enhancement must be established by research data rather than asserted.

The access dimension, which formed the moral background of the litigation, has a legal home of its own in the health statutes, and the relationship between patent status and the regulation of medicines is addressed in the complete guide to health laws in India and, for the marketing approval regime itself, in the article on the Drugs and Cosmetics Act 1940. This article notes the relationship only to fix a point the clause itself presupposes: a patent and a marketing approval are different instruments answering different questions, and the grant or refusal of the one neither confers nor removes the other.

The comparison other systems offer

Every patent system must answer the question the clause answers, namely what to do about the valuable but derivative variant, and the differences between the answers are structural rather than rhetorical, which is why comparison is instructive here. The United States has no equivalent of Section 3(d). A polymorph or salt claim is tested under the general law: novelty, and non-obviousness as developed after KSR, together with the written description and enablement requirements. The consequence is that the inquiry the Indian statute conducts under the heading of efficacy is conducted in the United States, if at all, under the heading of obviousness, where evidence of unexpected results can carry a claim, and where improved stability, processability or bioavailability may figure as such results. The forum and the burden differ accordingly: in the United States the fight is usually in infringement litigation over a granted patent presumed valid; in India the fight begins in examination and opposition, before grant, against a statutory deeming that the derivative is the same substance. Two jurisdictions in India’s broader region chose the Indian instrument rather than the American one. The Philippines, in the Universally Accessible Cheaper and Quality Medicines Act, 2008, wrote into its intellectual property code an exclusion modelled on Section 3(d): in the case of drugs and medicines, the mere discovery of a new form or new property of a known substance which does not result in the enhancement of the known efficacy of that substance is excluded from patent protection, with an Explanation that reproduces the Indian deeming list and the significant difference qualification almost word for word, and the same standard is written into the Philippine definition of inventive step as well. Argentina chose an administrative route to a similar destination: examination guidelines issued in 2012 directed examiners to treat claims to polymorphs, salts, formulations and analogous derivatives of known active ingredients as, in general, not patentable subject matter. Europe illustrates the opposite allocation of the same anxieties. The European Patent Convention contains no efficacy threshold for derivatives: a polymorph claim is examined for novelty and inventive step under the problem and solution approach, and a demonstrated improvement in stability or processability may support inventive step as a technical effect. The follow-on problem is managed in Europe principally outside patentability, through the regulatory exclusivities that protect clinical trial data and through supplementary protection certificates that extend, rather than multiply, the original right. The comparison isolates what is distinctive in the Indian design. India placed the filter inside the definition of patentable subject matter, at the front of the system, administered before grant on the applicant’s own documents; Europe placed its controls around the granted right and the regulatory dossier, after the patentability inquiry has run its ordinary course. The same commercial strategy meets a doctrinal wall in one system and a timetable in the other. The three systems share a premise the United States rejects: that the ordinary tests, administered case by case, under-deter follow-on filings in pharmaceuticals, because the applicant drafts with knowledge of the tests and the examiner sees one file at a time, while the strategy the tests must catch is visible only across a portfolio. Whether that premise is sound is the empirical dispute recorded in the previous section. Its adoption in Manila and Buenos Aires shows that the Indian provision is an exportable legislative technology and not a local eccentricity, whatever view is taken of its merits.

What the judgment requires of an applicant

The professional lesson of the case can be stated as four disciplines, each traceable to a specific failure in the record. The first is identification. An applicant for a new form must begin by identifying the known substance from which the form derives, because the Controller and the Court will identify it if the applicant does not, and the identification controls the comparison. Zimmermann teaches that a substance described in an earlier patent together with its salts is known, with its known efficacy, and that drafting the claim around the free base will not move the baseline. A corollary of identification is easily missed and worth stating separately, because in this case the decisive document was the applicant’s own. The Zimmermann patent was not a stranger’s publication unearthed by an opponent’s search: it was the work of the applicant’s own predecessor company, the foundation patent of the applicant’s own programme, and the specification from which the applicant’s entire development history descended. An applicant for a derivative claim must therefore audit its own portfolio before it audits the world’s. The earlier filings of the same house, with their genus claims, their salt language and their statements of pharmacological effect, are the first place an examiner or an opponent will look for the known substance and its known efficacy, and they are written, as the Zimmermann specification was written, in the generous idiom of a filing made when breadth cost nothing. The discipline the case imposes is retrospective candour: the drafter of the later application must read the earlier one as the Court will read it, identify the substance it made known, and build the efficacy case against that substance rather than against a predecessor chosen for convenience.

The second discipline is comparison. The efficacy case must be run against that identified substance, in the form immediately preceding the claim, not against a remote ancestor chosen because the data happens to exist. Novartis lost ground it could not recover by comparing against the free base and the alpha form when the law required comparison against the non-crystalline salt. The third discipline is the kind of evidence. Physical and processing advantages are discounted in advance by the Explanation; the dossier must speak to therapeutic effect, and a pharmacokinetic showing such as improved bioavailability must be connected by research data to a therapeutic benefit rather than left as an assumption. The fourth discipline is timing. The affidavits in Novartis were commissioned after the question arose, and the expert who filed them explained why: the study had never been done because, under the law as it stood when the application was prepared, there had been no occasion to do it. After 2013 there is always an occasion. The efficacy record belongs in the application as filed.

Two further practical consequences follow from the judgment’s allocation of the burden. The first concerns the contents of the specification as filed. Because the deeming in the Explanation places the burden on the applicant, and because the comparison must run against the known substance identified from the prior art, the specification for a derivative claim should itself identify that substance candidly, state the therapeutic respect in which the claimed form improves on it, and carry or reference the comparative data supporting the statement, rather than leaving the efficacy case to be constructed in reply to an examination report. Affidavit evidence in prosecution remains possible, as the Novartis record itself shows, but the affidavits there were handicapped by a specification that had framed the invention in purely physical terms, and they could not outrun the document they supplemented. The second consequence concerns claim drafting across a portfolio. A process claim stood outside the objection that defeated the product claim in this very litigation, and the Board allowed it; applicants who expect a Section 3(d) objection to a product claim for a new form therefore protect the manufacturing contribution separately, by process claims directed at the route of preparation, while the product claim is defended on the efficacy record. Neither consequence is a loophole. Both are the predictable professional adaptations to a threshold test administered on documents.

Readers applying this account to a live filing should treat it as legal information and not as advice on any specific application, and contested questions under the clause, particularly on the open points concerning safety and toxicity, belong with counsel instructed on the full specification and the full prior art.

The verdict

What is settled can be stated shortly. Section 3(d) is a second tier of qualifying standards, additional to novelty and inventive step, for new forms of known substances. The known substance is identified objectively from the prior disclosure, and where an earlier patent discloses a compound with its salts and its properties, the salt is the known substance and the comparison runs against it. Efficacy in a medicine means therapeutic efficacy, judged strictly and narrowly, and the properties inherent in a change of form do not count toward it. Enhancement must be claimed specifically and established by research data, and increased bioavailability does not by itself discharge that burden. The provision does not bar incremental invention as such, and neither the Supreme Court nor any WTO adjudicator has pronounced on its compatibility with TRIPS. What remains open is equally definite: whether improved safety or reduced toxicity can amount to enhanced therapeutic efficacy, whether efficacy bears the narrower pharmacodynamic meaning urged by the patients’ intervenor, and how the threshold character of the clause, affirmed by the Delhi High Court, orders its interaction with the definition of invention in contested litigation. The provision’s namable claim survives the scrutiny of its own record: under Indian law as construed in 2013, new and non-obvious is not enough. For a researcher or practitioner holding sources, data and correspondence on a contested provision of this kind, the filing discipline matters as much as the legal one, and the records of litigation and research are best kept in a private workspace of the kind described for source protection in the VaultBook workflow for journalists protecting their sources.

Frequently Asked Questions

Q: What does Section 3(d) actually say?

Section 3(d) of the Patents Act, 1970 provides that the mere discovery of a new form of a known substance which does not result in the enhancement of the known efficacy of that substance is not an invention. The clause also covers the mere discovery of any new property or new use for a known substance, and the mere use of a known process, machine or apparatus unless the process yields a new product or employs a new reactant. Its Explanation deems salts, esters, ethers, polymorphs, metabolites, pure form, particle size, isomers, mixtures of isomers, complexes, combinations and other derivatives of a known substance to be the same substance, unless they differ significantly in properties with regard to efficacy. The clause does not define efficacy, which is why the Supreme Court’s construction of the word in the Novartis case carries the weight it does.

Q: What did the Novartis judgment decide?

In Novartis AG v Union of India, (2013) 6 SCC 1, decided on 1 April 2013, the Supreme Court dismissed Novartis’s appeals and refused the patent for the beta crystalline form of imatinib mesylate, the cancer drug Glivec. The Court held that imatinib mesylate was itself a known substance from the Zimmermann patent of the 1990s, which had disclosed imatinib together with its salts and its anti-tumour properties, so that the beta form was a new form of a known substance within Section 3(d). It held that efficacy in the case of a medicine means therapeutic efficacy, judged strictly and narrowly, and that the flow properties, thermodynamic stability and lower hygroscopicity shown for the beta form were properties of the kind the Explanation excludes. It dismissed the appeals with costs and allowed the appeals of Natco Pharma and the Cancer Patients Aid Association against findings recorded in Novartis’s favour below.

Q: How do courts read efficacy under Section 3(d)?

The Supreme Court read efficacy by its ordinary meaning, the ability to produce a desired or intended result, and held that the test depends on the function, utility or purpose of the product claimed. For a medicine that claims to cure a disease, the test of efficacy can only be therapeutic efficacy, and given the circumstances in which the clause was made more constrictive in 2005, that therapeutic efficacy must be judged strictly and narrowly. Not every advantageous or beneficial property qualifies: only properties that directly relate to efficacy count, and the Explanation indicates what does not qualify, since properties inherent in a form, such as the solubility of a salt or the hygroscopicity of a polymorph, are not themselves an enhancement of efficacy. The Court recorded, without deciding, two competing refinements: that efficacy is strictly a pharmacodynamic property, and that improved safety or reduced toxicity might also bear on therapeutic efficacy.

Q: What does evergreening mean and how does the Act address it?

Evergreening describes the strategy of extending the effective monopoly over one medicine beyond the term of its original patent by obtaining successive patents on variants of the known substance: a salt, a polymorph, a formulation, a dosage form, a new use, each filed as the earlier right approaches expiry, so that generic competition is deferred although the treatment itself has not advanced. The Act addresses the practice at the threshold rather than at the term. Section 3(d) does not shorten any patent or disturb the original product patent. It provides that a variant of a known substance is not an invention at all unless it results in enhancement of the known efficacy of that substance, and its Explanation treats listed derivatives as the same substance unless they differ significantly in properties with regard to efficacy. A variant shown to treat better passes the filter and remains patentable. A variant shown only to handle, store or process better does not.

Q: Does Section 3(d) bar all incremental innovation?

No, and the Supreme Court said so in terms. The judgment records that it is not to say that Section 3(d) bars patent protection for all incremental inventions of chemical and pharmaceutical substances, and that it would be a grave mistake to read the judgment as meaning the clause was meant to undo the fundamental change brought about when Section 5 was deleted and product patents for medicines were introduced. The clause sets a qualifying standard, not a prohibition. An incremental invention that results in enhancement of the known efficacy of the substance, demonstrated by evidence, satisfies the standard and may be patented, subject to the ordinary requirements of novelty, inventive step and industrial application. What the clause bars is the patenting of a new form of a known substance on the strength of advantages that do not amount to enhanced efficacy, however useful those advantages may be in manufacture or distribution.

Q: How has Section 3(d) been applied since Novartis?

The judgment was carried into examination practice by the pharmaceutical examination guidelines issued by the Office of the Controller General in October 2014, which instruct that an incremental invention based on a known substance with established medicinal activity falls foul of patentability if it fails to demonstrate significantly improved therapeutic efficacy over the known compound. Controllers apply the clause as a standard objection to derivative claims, and applicants are expected to place comparative efficacy material in or with the specification rather than develop the case in correspondence. In litigation, a Division Bench of the Delhi High Court later treated Section 3 as a threshold condition of patent eligibility, applied before the invention inquiry, and the Erlotinib proceedings recorded a Controller’s refusal of a polymorph claim whose proved advantage, thermodynamic stability, was of the same physical character as the advantages rejected in Novartis. Appeals in patent matters lie to the High Courts following the abolition of the Appellate Board in 2021.

Q: Is Section 3(d) consistent with TRIPS?

The question has no adjudicated answer. Novartis raised non-compliance with TRIPS before the Madras High Court, which in 2007 declined to decide it, holding that conformity with the agreement is a matter for the dispute settlement system of the World Trade Organization rather than for a municipal court, and Novartis did not appeal that decision. The Supreme Court in 2013 made no finding of compliance or non-compliance. No member state has filed a complaint against the provision in the WTO system, so its consistency with the agreement remains undetermined in dispute settlement. The arguments are rehearsed in the literature and in trade forums such as the Special 301 process: one side relies on Article 27.1 and non-discrimination between fields of technology, the other on the absence of any treaty definition of invention and on the public health flexibilities confirmed by the Doha Declaration of 2001.

Q: What counts as the known substance for Section 3(d)?

The known substance is the substance identified objectively from the prior disclosure, and Novartis shows how the identification is made. The Court held that imatinib mesylate was a known substance from the Zimmermann patent itself, because that specification disclosed imatinib among its compounds, described the compounds together with their acid addition salts, explained that references to the free compounds included the corresponding salts where appropriate, and disclosed the pharmacological properties. A substance so disclosed is known together with its known efficacy, whether or not the particular salt was named. Counsel’s attempt to present the salt as a first invention, and the beta form as a second invention built on it, failed because the application itself spoke of the free base and the alpha form, and because the disclosure answered the question. Once the known substance was identified as imatinib mesylate, enhancement had to be shown over imatinib mesylate in non-crystalline form, the substance immediately preceding the claimed form.

Q: What must an applicant show, and with what evidence, to establish enhanced efficacy under Section 3(d)?

The applicant must show that the claimed form results in enhancement of the known efficacy of the known substance, and for a medicine that means an enhancement of therapeutic efficacy, judged strictly and narrowly. The showing must be made by research data, not by assertion: the Supreme Court held that whether an increase in bioavailability leads to an enhancement of therapeutic efficacy in any given case must be specifically claimed and established by research data, and it found the Glivec record wanting because there was nothing on the point apart from the submissions of counsel. The data must compare the claimed form against the known substance it derives from, and the comparison must be directed at therapeutic effect. Evidence of better flow, stability, storage or processing does not answer the question, because the Explanation treats such properties, being inherent in the form, as outside efficacy. In practice the material belongs in the application as filed, supported where necessary by expert evidence during prosecution.

Q: Does Section 3(d) apply to chemical substances other than medicines?

Yes. The Supreme Court recorded that the provision, as amended, was meant especially to deal with chemical substances, and more particularly pharmaceutical products, and its language is not confined to medicines: it speaks of a new form of a known substance and of the enhancement of the known efficacy of that substance, without limiting the class of substances covered. The Explanation’s list of derivatives, salts, esters, ethers, polymorphs, metabolites, isomers and the rest, is the vocabulary of chemistry generally. What changes outside the pharmaceutical field is the content of efficacy, because the test follows the function, utility or purpose of the product: for a medicine the desired result is therapeutic, while for another chemical product the desired result is whatever the product is for, such as the performance of an agrochemical against its target pest. The clause’s litigation history is dominated by pharmaceuticals because that is where product patents, mailbox applications and follow-on filings were concentrated after 2005.

Q: Can a new property or new use of a known substance ever lead to a patent in India?

The clause bars the mere discovery of any new property or new use for a known substance, so the discovery as such is not an invention, however valuable the property or use turns out to be. That does not exhaust the possibilities. A claim directed to something more than the discovery, such as a new product or a process that employs the newly found property to produce a new result, is assessed under the ordinary provisions of the Act rather than under this limb, and the third limb of the clause itself saves a known process that results in a new product or employs at least one new reactant. The Act also removes certain use claims from this clause’s neighbourhood altogether: a method of treatment of human beings or animals is excluded by Section 3(i), and a substance obtained by mere admixture is excluded by Section 3(e). The practical distinction is between claiming the discovery and claiming an invention built on it; the first is barred by the clause, the second stands or falls on the general law.

Q: How does the Explanation to Section 3(d) treat salts, polymorphs, metabolites and other derivatives?

The Explanation deems salts, esters, ethers, polymorphs, metabolites, pure form, particle size, isomers, mixtures of isomers, complexes, combinations and other derivatives of a known substance to be the same substance as the known substance, unless they differ significantly in properties with regard to efficacy. Two consequences follow. The derivative begins as the same substance in law, so the applicant cannot treat the change of form as itself an invention and must displace the statutory identity by proof. And the proof required is specific: the difference must be significant, and it must be a difference in properties with regard to efficacy, which for a medicine means therapeutic efficacy. The Supreme Court added that the Explanation indicates what is not efficacy: each listed form has properties inherent to it, solubility in the case of a salt and hygroscopic behaviour in the case of a polymorph, and producing only the properties inherent in the new form does not amount to enhancement.

Q: What role did the Zimmermann patent play in the Novartis case?

The Zimmermann patent, United States Patent No 5,521,184, granted in 1996 on the work of Jurg Zimmermann and colleagues at Ciba-Geigy, was the foundation of the decision. It disclosed the family of compounds to which imatinib belongs, included imatinib as one of its thirty seven worked examples, described the compounds together with their acid addition salts and explained that references to the free compounds included the salts, and disclosed the anti-tumour and pharmacological properties of what it claimed. The Assistant Controller relied on it for anticipation and obviousness, and the Appellate Board reversed him on both. In the Supreme Court it performed a different function: it established what was known. Because the mesylate salt and its efficacy were within its disclosure, imatinib mesylate was the known substance for Section 3(d), and Novartis’s argument, drawn from the United States decision in Hogan, that a genus patent does not cover later species, was rejected on the objective material.

Q: What did the Supreme Court say about the price of Glivec and the generic versions sold in India?

Very little, and the little it said was quotation rather than holding. The judgment records the Appellate Board’s discussion, in which the Board noted that during the period of exclusive marketing rights Novartis charged Rs 1,20,000 per month for a required dose of Glivec, a figure the Board described as not disputed by the company, and on that footing questioned, under Section 3(b), whether the price level made the grant injurious to public health. The Supreme Court reproduced that passage and observed that the Board had additionally held the patent barred by Section 3(b) for that reason. The Supreme Court’s own decision does not rest on price or on Section 3(b) at all: it rests on the construction of Section 3(d) and on the absence of evidence of enhanced therapeutic efficacy. Accounts that present the judgment as a decision about drug pricing misstate its reasoning, although the price difference between the branded product and generic versions framed the public importance of the case.

Q: Against what baseline is enhancement of efficacy measured under Section 3(d)?

Enhancement is measured against the known efficacy of the known substance from which the claimed form derives, identified objectively from the prior disclosure, and in practice against the form of that substance immediately preceding the claim. In Novartis the claimed beta crystalline form derived from imatinib mesylate, so the comparison required was between the beta form and imatinib mesylate in non-crystalline form. The comparisons the applicant actually offered failed that discipline. The physical advantages, flow, thermodynamic stability and hygroscopicity, were asserted against the alpha crystalline form, and the bioavailability study compared the beta form against imatinib free base. There was no material comparing efficacy, or even solubility, between the beta form and non-crystalline imatinib mesylate, and the Court noted that the higher solubility relied on might be a property of the salt itself rather than of the beta form. Choosing the wrong baseline left the statutory comparison unmade, and the application failed on that ground independently of the meaning of efficacy.

Q: What is a selection patent, and how does Section 3(d) bear on the choice between genus and species claims?

A selection patent claims one member, or a small group, from a larger class already disclosed in an earlier patent, on the footing that the selected member has advantages the class as a whole was not known to possess. The tension is structural: the earlier disclosure may be read broadly, as covering everything within the genus including the species, or narrowly, as disclosing the genus without making any individual species known. Novartis presented the problem in salt form. Its counsel described a first invention selecting imatinib and converting it into the mesylate salt, and a second invention arriving at the beta crystalline form, and relied on the United States decision in Hogan for the proposition that a genus disclosure does not anticipate a later species. The Court declined to adopt that approach and answered on the documents: the Zimmermann specification disclosed the compounds with their salts and their properties, so the salt was known, and the scope of a patent should not be determined by the artful drafting of its claims. Under Section 3(d), a selected derivative must additionally show enhanced efficacy over the substance it was selected from.

Q: At which stages can Section 3(d) be raised: examination, opposition, or after grant?

At each of them. The clause is applied first in examination, where the Controller considers whether the claimed subject matter is an invention at all, and the pharmaceutical examination guidelines direct examiners to test derivative claims against the enhanced therapeutic efficacy standard. It is a ground of opposition both before grant and after grant under Section 25, since an opponent may contend that the invention claimed is not an invention within the meaning of the Act; the Novartis litigation itself began as pre-grant oppositions filed by a patients’ association and four generic companies. After grant, the same contention is available in revocation proceedings under Section 64 and as a defence in infringement litigation, where the court examines patentability for itself. The clause therefore operates as a continuing condition on the right rather than a single hurdle at filing. A patent granted over a Section 3(d) objection, or without the objection having been pressed, remains exposed to the same test whenever its validity is challenged in court.

Q: How does Section 3(d) differ from Section 3(e), the exclusion of mere admixtures?

The two clauses police different strategies and ask different questions. Section 3(e) excludes a substance obtained by a mere admixture resulting only in the aggregation of the properties of its components, together with a process for producing such a substance: its target is the combination whose whole is no more than the sum of its parts, and its test is whether the mixture shows any property beyond aggregation, such as a synergistic effect. Section 3(d) is not about combinations at all. It addresses a single known substance in a new form, or a new property or use of it, and its test is enhancement of the known efficacy of that substance. A formulation may have to pass both: the claimed form of the active ingredient is tested under clause (d), and the combination of ingredients is tested under clause (e), and neither inquiry answers the other. Clause (d) looks backward to a known substance and demands improvement over it; clause (e) looks inward at the components of a mixture and demands more than their addition.

Q: Does getting drug approval in India depend on the patent position under Section 3(d)?

No. The patent system and the drug approval system are separate regimes answering separate questions. A patent is granted or refused by the Patent Office under the Patents Act, and Section 3(d) belongs to that inquiry alone: it decides whether a claimed form is an invention, not whether a drug may be sold. Approval to manufacture and market a drug is granted under the Drugs and Cosmetics Act, 1940 and its rules, on material directed at quality, safety and efficacy in the regulatory sense, and India does not operate a system of patent linkage under which the drug regulator withholds approval because a patent exists or because a patent application is pending. A generic manufacturer may therefore hold a marketing approval for a product whose patent status is disputed, and the patent holder’s remedy lies in infringement proceedings in court rather than before the regulator. The refusal of the Glivec patent removed a barrier to competition; it did not itself grant or remove any approval.

Q: Why does Section 3(d) describe its subject as a mere discovery rather than an invention?

Because the Act draws a line between discovering what exists and inventing what does not, and Section 3 polices that line throughout. Section 3 opens by declaring that the listed matters are not inventions within the meaning of the Act, and its neighbouring clauses exclude the discovery of scientific principles and of substances occurring in nature. The word mere does the operative work in clause (d). A new form of a known substance is treated as something found rather than made: the substance and its efficacy already exist, and isolating a crystalline state of it is presented by the statute as an act of recognition. The clause then supplies the route by which such a subject escapes the description: if the new form results in enhancement of the known efficacy, the claim is no longer a mere discovery and the clause does not bite. The drafting thus converts a philosophical distinction into an evidentiary one. The applicant who shows enhanced therapeutic efficacy by research data is treated as an inventor; the applicant who shows only the form is treated as a discoverer.