US-headquartered pharmaceutical companies sourcing active pharmaceutical ingredients from Indian or Chinese manufacturers have long worked from a familiar quality-assurance map. The active-substance manufacturer is qualified under cGMP per 21 C.F.R. Parts 210 and 211; the supplier-quality agreement carries cGMP audit rights and inspection-cooperation undertakings; the FDA conducts pre-approval and surveillance inspections at the foreign manufacturing site through its foreign-inspection programme. The EU Pharma Package, the most consequential pharmaceutical reform the Union has undertaken in two decades, introduces an obligation that has no analogue in that map.
The Environmental Risk Assessment that must accompany every marketing-authorisation application is being expanded, for antimicrobial products, to cover the entire manufacturing supply chain inside and outside the Union, with antimicrobial-resistance selection in the environment treated as a distinct analytical limb. The reach is not the FDA framework extended to environmental discharge; it is a structurally different instrument that operates on different inputs (effluent characterisation and production-volume data from the API site rather than process-control records from the production envelope), against different thresholds (in-environment resistance-selection concentrations rather than batch-conformity specifications), and with different consequences (refusal, suspension, or revocation of the marketing authorisation rather than batch rejection or facility warning). The data that disclosure obligation will require sits at the API production site abroad, not inside any quality file the US sponsor holds under its existing supplier arrangements. The new framework does not arrive immediately: under Art. 219 of the Directive, Member States must transpose the new code within 18-to-24 months of entry into force and apply its provisions from 24 months after entry into force, with publication in the Official Journal still pending as of this writing. The contractual architecture the obligation will require, however, is the kind that takes longer to renegotiate than the transitional window allows.
1. The ERA Before and After: Scope, Trigger, and Consequence
The Environmental Risk Assessment has been a formal component of the EU marketing-authorisation dossier since 2004, when the predecessor regime was inserted into Art. 8(3)(ca) of Directive 2001/83/EC.1Directive 2001/83/EC, the EU code on medicinal products for human use, in force until repealed by the Pharma Package Directive. Its scope was modest. The operative EMA scientific guideline, first issued in 2006 and revised into Revision 1 effective 1 September 2024, anchored the assessment in the use-and-disposal phase of the product's lifecycle, with predicted environmental concentrations derived from estimated patient consumption and excretion rates measured against predicted no-effect concentrations for aquatic organisms.2EMA scientific guideline on the environmental risk assessment of medicinal products for human use, first issued 2006 and Revision 1 effective 1 September 2024. Revision 1 expanded technical guidance on tailored assessments for antibacterials and endocrine-active substances and absorbed the questions-and-answers practice that had accumulated since 2006, but it did not itself create the manufacturing-supply-chain AMR limb that arrives only with Art. 22 of the new Directive. Manufacturing-phase emissions were almost entirely outside the analytical frame. Antimicrobial-resistance selection was not a separate ERA limb. And the assessment carried no marketing-authorisation-level consequence at all: for human medicinal products, the environmental impact, while it had to be assessed and submitted, could not constitute a criterion for refusal of the authorisation. A defective ERA generated clock-stop delays at validation; it could not, by itself, sink the application.
The Pharma Package rewrites all three dimensions. Art. 22 of the new Directive lifts the ERA out of an annex-driven format and into a stand-alone provision with five operative paragraphs, and the burden it imposes is not uniform across the product universe. Two of the enhancements apply to all medicinal products; one applies only to antimicrobials. For every medicinal product, the new Directive requires the assessment to indicate whether the product or any of its ingredients meets the criteria for classification as persistent, bioaccumulative, and toxic (PBT); very persistent and very bioaccumulative (vPvB); persistent, mobile, and toxic (PMT); very persistent and very mobile (vPvM); or endocrine-active, by reference to the criteria in Annex I to Regulation (EC) No 1272/2008 on classification, labelling, and packaging of substances and mixtures.3Council compromise text for the Pharma Package Directive, 24 February 2026; Art. 22 of the Directive contains the operative ERA provisions.4Regulation (EC) No 1272/2008 on classification, labelling and packaging of substances and mixtures, the CLP Regulation that supplies the hazard criteria. Also for every medicinal product, risk-mitigation measures must be proposed where the product or its ingredients fall within the pollutant lists of the Water Framework Directive 2000/60/EC, the Environmental Quality Standards Directive 2008/105/EC, the Groundwater Directive 2006/118/EC, or the Industrial Emissions Directive 2010/75/EU, with a duly substantiated demonstration that the proposed measures are appropriate and sufficient.5The four EU environmental directives cross-referenced in Art. 22(3) of the Directive as supplying the pollutant lists. The third enhancement is the one that opens the supply-chain dimension and is the focus of this article: for antimicrobials only, Art. 22(4) of the Directive requires the ERA to include an evaluation of the risk for antimicrobial-resistance selection in the environment due to the entire manufacturing supply chain inside and outside the Union, alongside use and disposal, taking into account international standards on antibiotic-specific predicted-no-effect concentrations. The category is broader than antibiotics in the strict sense; Art. 22(5) of the Directive expressly contemplates "antimicrobials other than antibiotics," and the antimicrobial-resistance definition in Art. 4(34) of the Directive operates across micro-organism classes, so antifungals, antivirals, and antiparasitics fall within the operative scope, even if antibacterials remain the dominant practical case. The reach is therefore not confined to legacy antibiotic franchises: large-volume US-sponsor antiviral portfolios (the HIV and hepatitis-C franchises with established Indian or Chinese active-substance sourcing) and the major systemic-antifungal portfolios face the same supply-chain obligation under Art. 22(4) of the Directive, even though the AMR pathway for non-antibiotic antimicrobials is methodologically less developed than for antibiotics.
The ERA has been reorganised from a documentation exercise calibrated to the patient's medicine cabinet and the wastewater plant into a supply-chain disclosure obligation calibrated to the API production site abroad, with marketing-authorisation-level consequences for non-performance.
The consequence escalator has been rewritten in parallel. Under Art. 47(1)(d) of the new Directive, a national competent authority shall refuse the marketing authorisation if the ERA is incomplete or insufficiently substantiated by the applicant, or if the risks identified in the assessment have not been sufficiently addressed including through risk-mitigation measures, unless the applicant has duly justified the position and the authority considers that the marketing authorisation can be granted subject to post-authorisation conditions.3 Once a marketing authorisation is in force, Art. 195(2) of the Directive permits the competent authority (or, for centralised authorisations, the Commission) to suspend, revoke, or vary the authorisation where a serious risk to the environment or public health has been identified and not sufficiently addressed by the marketing-authorisation holder, and Art. 196(1)(f) of the Directive triggers a parallel supply prohibition and market-withdrawal mechanism on the same ground. The pre-Pharma-Package position, in which ERA defects were procedurally annoying but commercially containable, no longer holds. The Pharma Package Regulation reinforces the architecture from the centralised-authorisation side: under Art. 12(4)(j) of the Regulation, the Committee for Medicinal Products for Human Use may attach to a favourable opinion the obligation to conduct post-authorisation ERA studies, collect monitoring data, or implement risk-mitigation measures where concerns about risks to the environment or public health, including antimicrobial resistance, remain to be investigated.6Council compromise text for the Pharma Package Regulation, 24 February 2026; Art. 12(4)(j) of the Regulation supplies the centralised committee's post-authorisation ERA lever.
2. Reaching the Manufacturing Footprint Outside the EU
The textual hook for the extraterritorial reach is short. Art. 22(4) of the Directive requires the AMR limb of the ERA to evaluate resistance-selection risk due to "the entire manufacturing supply chain inside and outside the Union, use and disposal, including by healthcare professionals and patients" of the antimicrobial in question.3 The policy framing is supplied by three recitals reading in series. Recital 2a anchors the whole Directive in the One Health Approach and identifies "pollution from active pharmaceutical ingredients" as a cause of antimicrobial resistance increasing "rapidly, posing risks to public health globally." Recital 65a operationalises the same point in the AMR section, calling for coordinated action throughout the supply chain. Recital 72 then closes the textual ground for the extraterritorial reach: emissions and discharges of antimicrobials from manufacturing sites are "a global concern regardless where the emissions and discharges take place," and the ERA scope should therefore be extended to cover AMR selection during the entire life cycle including manufacturing.
The legal mechanism by which the EU framework reaches the Asian production site is not direct. The EU has not asserted regulatory jurisdiction over the Indian or Chinese API manufacturer as such. The reach operates through the EU marketing-authorisation holder, who is the entity submitting the ERA, the entity facing refusal under Art. 47(1)(d) of the Directive if the assessment is incomplete, and the entity facing suspension or revocation under Art. 195(2) of the Directive if the risks remain unaddressed once the authorisation is in force. The EU marketing-authorisation holder must produce the data; the question is whether the contractual architecture between that holder and the upstream manufacturer can compel it. That question has a different answer depending on which strand of EU regulation is asked. The GMP-audit infrastructure under Art. 147(1)(f) and (g) of the new Directive requires manufacturing-authorisation holders to use only active substances manufactured in accordance with GMP for active substances and to verify compliance by conducting regular audits at the manufacturing and distribution sites of the API manufacturer.3 That infrastructure has a 20-year operational track record reaching into Indian and Chinese API sites. It is not, however, designed to generate environmental-effluent data. A GMP audit inspects process control, batch records, change management, deviation handling, and contamination control inside the production envelope. It does not sample the receiving water body downstream of the plant, nor does it generate the antibiotic-effluent concentration data that an AMR-ERA limb requires.
The asymmetry between what the audit infrastructure inspects and what the ERA limb requires sits at the centre of the supply-chain reach. The new Directive obliges the EU marketing-authorisation holder to deliver effluent-grade environmental data that the existing audit programme was never built to collect. The audit-right clauses of the standard active-substance supply agreement, drafted to the cGMP and EU GMP comparator and refined over two decades of practice, do not on their face reach effluent sampling at the production site, nor measurement of antibiotic concentrations in receiving water bodies, nor disclosure of the production volume from which the predicted environmental concentration must be back-calculated. The US-anchored counterpart situation is structurally different. FDA's environmental-assessment framework under 21 C.F.R. Part 25 operates a categorical-exclusion regime that excludes most marketing applications from a substantive environmental review, with the assessment in the residual cases focused predominantly on use-phase release and disposal rather than on manufacturing-supply-chain effluent.721 C.F.R. Part 25, the FDA implementing regulation for the National Environmental Policy Act, supplying the US comparator. FDA practice has not developed a systematic, AMR-specific assessment of overseas manufacturing effluent comparable to the limb under Art. 22(4) of the Directive, and the FDA's enforcement reach into Asian manufacturing sites runs through cGMP inspection rather than through any instrument analogous to the AMR-ERA limb.
3. AMR-Specific ERA: Resistance Selection in the Environment as a Separate Limb
Treating the AMR limb as a doctrinally separate analytical step, rather than as a sub-question inside the general ERA, is the conceptual move that does the work. The definition of "environmental risk assessment" in the Pharma Package Directive itself signals the bifurcation: for medicinal products with an antimicrobial mode of action, the assessment "also encompasses an evaluation of the risk for antimicrobial resistance selection in the environment due to the manufacturing, use and disposal of that medicinal product."3 The general ERA limb evaluates ecotoxic effects on aquatic and terrestrial organisms; the AMR limb evaluates a different endpoint, the in-environment selective pressure that drives bacterial-resistance evolution. The two limbs share input data and analytical infrastructure, but they answer different scientific questions and they read against different thresholds. Art. 22(4) of the Directive anchors the AMR limb in international standards on predicted no-effect concentrations specific to antibiotics, where such standards exist; for antimicrobials other than antibiotics, the EMA is to issue scientific guidelines under Art. 22(5) of the Directive, consulting the European Chemicals Agency, the European Food Safety Authority, the European Environment Agency, and the European Centre for Disease Prevention and Control. The wording "international standards" carries a degree of slippage worth flagging: as of the date of this article, no codified multilateral PNEC annex exists. The de facto reference points are industry-developed lists, most prominently the AMR Industry Alliance's publicly maintained antibiotic-PNEC schedule, alongside academic and consensus literature and OECD and WHO work-streams that have not yet crystallised into binding norms.
The empirical foundation for the manufacturing-supply-chain reach is not abstract. Independent monitoring of effluent from pharmaceutical-cluster wastewater treatment plants serving bulk-drug manufacturers, most prominently the Larsson and colleagues 2007 study of the Patancheru common effluent treatment plant near Hyderabad that serves around 90 active-substance manufacturers, found ciprofloxacin concentrations in plant outflow at up to 31,000 micrograms per litre, more than a thousand-fold above concentrations known to be toxic to bacteria and very far above any plausible predicted no-effect concentration for resistance selection.8Larsson et al, ‘Effluent from drug manufactures contains extremely high levels of pharmaceuticals’ (2007) 148 J Hazard Mater 751. Comparable measurements at Indian and Chinese clusters in the subsequent decade established the same pattern at additional sites and across multiple antibiotic classes. The policy decision in Art. 22(4) of the Directive to bring manufacturing emissions into the ERA frame rests on this empirical record. The legislative move is not a precautionary speculation about possible AMR selection at sites that may discharge antibiotics; it is a regulatory response to a measured fact about sites that have discharged them.
The verification question that the new framework does not resolve is who measures what at the third-country production site. The text of Art. 22 of the Directive places the obligation on the marketing-authorisation applicant; it does not contemplate an Agency inspector flying to the production cluster with sampling bottles. The pre-Pharma-Package ecotoxicity workflow operated on calculation rather than on measurement: predicted environmental concentrations were derived from consumption-and-excretion modelling, and predicted no-effect concentrations were derived from controlled laboratory studies. That calculation infrastructure can be carried into the AMR limb at the use-and-disposal end of the lifecycle, but it does not transpose cleanly onto manufacturing-effluent disclosure, where the relevant variables are site-specific production volumes, on-site wastewater treatment performance, and receiving-water-body dilution factors that only the API producer can supply. Whether competent authorities will accept self-declaration from the third-country supplier, third-party audit reports against an emerging effluent-disclosure standard, or only on-site inspection through bilateral arrangements with the producing state's environmental regulator, is a question the Directive leaves to the EMA scientific guideline that has not yet issued.
The Directive does, however, point toward a portfolio-level architecture that will matter operationally for sponsors with multiple products sharing an API. Art. 24 of the Directive mandates an active-substance-based ERA-monograph system run by the EMA in cooperation with national competent authorities, with risk-based prioritisation of substances, a proof-of-concept pilot to be completed within three years of the Directive's entry into force, and delegated acts to specify the content, format, and use of monographs in subsequent marketing-authorisation applications.3 Once the monograph for a given active substance exists, downstream sponsors relying on the same API should in principle be able to refer to it rather than reconstruct the assessment from scratch. The pilot's three-year horizon and the delegated-act build-out mean the monograph system will not be operational for the first wave of post-application ERAs; sponsors filing in the early years of the new regime will be assembling the data the monograph system later consolidates. Whether the data they assemble survives consolidation, or has to be regenerated to a methodology the EMA only finalises later, is the kind of asymmetric-investment question that does not have a clean answer at filing time. A further cross-statutory wrinkle arises where the active substance also falls within the scope of REACH (Regulation (EC) No 1907/2006) or the Biocidal Products Regulation (Regulation (EU) No 528/2012); some antimicrobials (cetrimide and chlorhexidine being the standard examples) sit awkwardly across the pharma/biocide line and may carry parallel registration or assessment obligations whose underlying datasets overlap with the ERA. Recital 71 of the Directive expressly flags the REACH, Biocidal, Pesticide and Veterinary-Medicines frameworks as parallel risk-assessment regimes for medicinal-product chemicals, but the coordination among them is to be developed rather than already specified.
The interaction with the existing manufacturing-and-distribution-supply-chain framework illustrates how partial the answer is. The qualified person obligations under Art. 153 of the new Directive require batch-release certification for medicinal products imported from third countries, with full qualitative analysis, quantitative analysis of the active substances, and the necessary controls to ensure conformity with the marketing authorisation.3 That regime governs whether a particular batch can be released to the EU market; it does not govern whether the cumulative manufacturing effluent at the API site has produced a selective pressure consistent with the resistance assumptions in the dossier. The two regimes operate at different time horizons, on different evidentiary inputs, and against different consequence triggers. A US sponsor whose API supply chain has been operationally satisfactory under the batch-release regime for twenty years cannot infer from that record anything about how its AMR-ERA limb will be assessed.
4. Denial and Revocation as Consequences: The MA-Level Stakes
The pre-Pharma-Package position was that an ERA deficiency surfaced primarily at the validation stage of the marketing-authorisation procedure. The competent authority sent a request for information, the clock stopped, the applicant responded, the dossier resumed. Refusal of the marketing authorisation on environmental grounds was, for human medicinal products, not available at all under the predecessor regime: the environmental impact had to be assessed and submitted, but its outcome could not constitute a criterion for refusal of the authorisation, and revocation on environmental grounds did not arise. The Pharma Package removes that carve-out. The new framework moves the consequence locus upward through the procedure and outward across the lifecycle. Refusal is an explicit, freestanding ground under Art. 47(1)(d) of the new Directive: the marketing authorisation shall be refused where the ERA is incomplete or insufficiently substantiated, or where the risks identified in the assessment have not been sufficiently addressed, unless the applicant has duly justified the position and the competent authority concludes that the authorisation can be granted subject to post-authorisation conditions to conduct further ERA studies or implement risk-mitigation measures.3
Post-grant, the consequence escalator continues. Under Art. 195(2) of the Directive, the competent authority of the Member State (or, in the case of centralised authorisations, the Commission) may suspend, revoke, or vary a marketing authorisation where a serious risk to the environment or public health has been identified and not sufficiently addressed by the marketing-authorisation holder. The wording does not require demonstration of patient-level harm; it requires identification of a serious environmental risk that the holder has not adequately mitigated. Recital 34c of the Directive reinforces the same point through the conditions-attached-to-the-authorisation route, expressly noting that competent authorities should be able to act where conditions attached to the authorisation, including the condition to conduct post-authorisation ERA studies, are not complied with. Art. 196(1)(f) of the Directive supplies the parallel supply-prohibition and market-withdrawal mechanism on a closely cognate ground (a serious risk to the environment, or to public health via the environment), with the difference that withdrawal operates batch-level if the competent authority limits the prohibition to specific batches under Art. 196(2) of the Directive.
The in-life trigger that connects the two regimes is the update obligation in Art. 22(6) of the Directive, read with the variation regime in Art. 90 of the Directive. Art. 22(6) of the Directive requires the marketing-authorisation holder to update the ERA without undue delay if new information relevant to the assessment criteria becomes available and could change the conclusions, including monitoring data and updated risk assessments under other Union legislation. Art. 90 of the Directive then channels material changes through the variation procedure. The structural consequence is that a post-grant deterioration in the supplier-effluent profile (for example, new monitoring data showing antibiotic concentrations downstream of an Indian API site exceeding the resistance-selection PNEC assumptions in the dossier) is not merely a compliance fact for the holder to record. It is a variation trigger, and where the holder cannot demonstrate that the resulting risk has been sufficiently addressed, it becomes a route into the suspension-or-revocation procedure under Art. 195(2) of the Directive. The application-stage scrutiny under Art. 47(1)(d) of the Directive and the lifecycle scrutiny under Art. 22(6), 90, and 195(2) of the Directive are not parallel tracks; they are stages of a single continuous obligation that the marketing-authorisation holder carries from filing through the duration of the authorisation.
The centralised-procedure pathway and the national-procedure pathway carry the same substantive grounds but different procedural choreography. For centralised authorisations granted under the Pharma Package Regulation, the Committee for Medicinal Products for Human Use's opinion under Art. 12(4)(j) of the Regulation can attach post-authorisation ERA obligations at the opinion stage; once granted, the Commission acts on subsequent suspension or revocation. For national authorisations granted under the Directive procedures, the competent authority of the granting Member State acts. The substantive ERA wording is shared across the two regimes, but the appeal architecture, the timeline for the suspension procedure, and the Member-State-versus-Commission decisional locus differ. For a US sponsor with a product authorised through the centralised procedure, the relevant interlocutor on an ERA-defect escalation is the EMA and ultimately the Commission; for a product authorised nationally in two or three Member States through the decentralised or mutual-recognition routes, the interlocutor is the reference Member State's competent authority, with potential referral to the coordination group under Art. 37 of the Directive if Member States diverge on the ERA assessment.3 The practical weight tilts toward the centralised pathway: most new antimicrobials are routed to the centralised procedure under the mandatory-centralised-product list in the Regulation's Annex, so for the products to which Art. 22(4) of the Directive most directly applies, the operationally relevant escalation interlocutor is the EMA and the Commission rather than any single national authority.
The discussion so far has focused on new marketing-authorisation applications, but the most significant exposure for sponsors with established antibiotic franchises sits elsewhere. Art. 23 of the Directive requires the EMA, within 30 months of entry into force and after consultation with national competent authorities, the ECDC, ECHA, EFSA, and the EEA, to establish a risk-based prioritisation programme requiring marketing-authorisation holders of medicinal products authorised before 30 October 2005 (which were never subject to any ERA at original authorisation) to submit ERAs "in accordance with Article 22" where the EMA identifies them as potentially harmful to the environment.3 The "in accordance with Article 22" cross-reference carries the AMR-supply-chain limb into the retroactive programme: pre-2005-authorised antimicrobials, including many of the most-prescribed antibiotics in routine clinical use whose first authorisations predate the original ERA requirement, are caught. The portfolio significance for US sponsors with legacy antibiotic franchises sourced through long-standing relationships with Indian or Chinese API suppliers is potentially larger than the new-MA exposure that has so far attracted attention, because the contracts that govern those legacy supplier relationships were drafted before any version of an environmental-disclosure obligation existed, and the retroactive submission obligation lands on the marketing-authorisation holder regardless of when the underlying authorisation was granted.
The structural tension that an antimicrobial-MA suspension or revocation on environmental grounds creates with the EU's parallel critical-medicines-security architecture deserves attention. Regulation (EU) 2022/123 already gives the EMA an expanded shortage-monitoring and crisis-management mandate, and the Commission's proposal for a Critical Medicines Act published in March 2025 contemplates further obligations on supply continuity for medicines identified as critical, with antimicrobials a prominent category.9Regulation (EU) 2022/123 on a reinforced role for the EMA in crisis preparedness, plus the Commission's March 2025 proposal for a Critical Medicines Act. A suspension under Art. 195(2) of the Directive triggered by serious environmental risk at an upstream API site, applied to an antibiotic on a Member State's critical-medicines list, would simultaneously satisfy one EU regulatory imperative and breach another. The Directive does not resolve the priority between the two regimes; the Critical Medicines Act proposal as of its March 2025 form does not clearly address the environmental-supply-disruption case. Whether competent authorities will, in practice, hold the suspension in abeyance pending an alternative qualified API source coming online, accept post-authorisation conditions in lieu of suspension, or proceed with the suspension and force the shortage-management apparatus to bridge supply, is a question the texts leave to enforcement discretion that has no track record yet.
5. Strategic Considerations: Supply-Chain Disclosure, Supplier-Contract Flowdowns, and the FDA cGMP Comparator
How much supply-chain disclosure a US sponsor can compel from an Asian active-substance supplier whose existing contract was drafted to FDA cGMP expectations is rarely a question that has been put to the contract before. The standard active-substance supply agreement carries audit rights against the GMP fact pattern, indemnities for batch nonconformity, and change-control notification undertakings keyed to the FDA Drug Master File. None of those mechanisms, in their typical form, reaches the effluent-disclosure or AMR-PNEC data that Art. 22(4) of the Directive contemplates. A US sponsor reading the new Directive against its existing supplier file will frequently find that the file contains no commitment from the upstream manufacturer to disclose production volumes by site, to characterise on-site wastewater treatment performance, or to make available the receiving-water-body data from which the predicted environmental concentration must be back-calculated. Whether the supplier is contractually obliged to provide that information, on what timeline, and to which standard of substantiation, will turn on contract language that was not drafted to anticipate the question. The standard supplier objection that detailed effluent and production-volume data is commercially confidential and cannot be shared on a publicly visible basis is partially answered by Art. 29(5) of the Directive, which requires competent authorities to delete information of a commercially confidential nature from the publicly released summary of the ERA assessment; but the protection is for the public summary, not for the underlying submission to the competent authority, and the contractual conversation with the upstream supplier has to absorb that distinction.3
Where the upstream Indian intermediate manufacturer simply refuses to provide environmental-effluent data, who bears the consequence inside the EU regulatory architecture is unambiguous on the regulatory side. The EU marketing-authorisation holder is the entity facing refusal under Art. 47(1)(d) of the Directive at the application stage, suspension or revocation under Art. 195(2) of the Directive post-grant, and supply prohibition under Art. 196(1)(f) of the Directive on the same environmental-risk ground. The upstream API manufacturer has no privity with the EU regulator on the ERA limb; its exposure runs through the supply contract back to the EU marketing-authorisation holder, where the typical force-majeure and exclusion-of-consequential-damages clauses were not designed to absorb the loss of an EU marketing authorisation. The contractual-allocation problem that emerges, separately analysed in this firm's earlier piece on pharma supply-chain liability where the regulatory consequence cannot be allocated by contract, compounds here: even where the supply agreement purports to push the loss back upstream, the regulator does not recognise the contractual allocation and continues to pursue the marketing-authorisation holder. Whether the existing contractual architecture distributes that loss in a way that any party would accept on a clear-eyed reading, particularly where the API is sourced from a single supplier with no qualified alternative, is one of the analytical questions a sponsor cannot answer from inside its own files.
The instinctive view that the FDA cGMP comparator gives the US sponsor leverage to compel the same data from the existing supplier base does not survive scrutiny. The supplier is already subject to FDA inspection, the supply agreement carries cGMP audit rights, the operational discipline is in place; yet FDA cGMP under 21 C.F.R. Parts 210 and 211 governs manufacturing process, not environmental discharge. FDA's environmental-assessment framework under 21 C.F.R. Part 25 operates a categorical-exclusion regime that excludes the great majority of marketing applications from substantive environmental review and, in the residual cases, focuses on use-phase release rather than manufacturing-supply-chain effluent. The US sponsor that approaches its Indian supplier with an audit request for effluent data, citing the cGMP audit-rights clause, is not making a request the existing contract supports. It is requesting a new category of disclosure under a clause designed for a different category of inspection, and the supplier's refusal is a contractually plausible position.
Closing the gap between what the existing agreement secures and what the obligation requires is not a single-clause amendment. It reaches the scope of the audit right, the basis on which the indemnity is sized, whether the sponsor can substitute an alternative qualified source if the incumbent supplier cannot meet the disclosure standard, and whether any cooperation duty survives long enough to track the life of the authorisation rather than the term of the supply contract. Each of those is a separately negotiated commercial term, and each sits in a part of the agreement that was drafted for a different purpose. Whether the gap can be closed at all in a renegotiation with an existing single-source API supplier, and at what counterparty cost, is a question the sponsor's commercial and legal functions must answer together. The retroactive posture is the harder problem: contracts negotiated five or ten years ago against a different regulatory baseline rarely reopen on cooperative terms when one side carries the EU-MA exposure and the other side does not.
A question that sits underneath the others, particularly for Swiss-headquartered groups operating into the EU through a subsidiary or through an EU-authorised representative arrangement, is how the chain of legal responsibility distributes intra-group exposure when the parent is itself a third-country entity. The chain is straightforward in outline. The Swiss parent sits contractually upstream of the EU marketing-authorisation holder, and the Asian API manufacturer sits contractually upstream of the Swiss parent. The AMR-ERA limb obligation runs through all of them, but for regulatory enforcement purposes it lands on the EU marketing-authorisation holder alone. Whether the intra-group contractual architecture distributes the resulting exposure in a manner that survives audit by the relevant Member State competent authority, or whether the intra-group arrangements were drafted to a quality-system fact pattern rather than to a regulatory-risk-allocation fact pattern, is the kind of question that surfaces only when the consequence is concretely in play. The US-sponsor framing is deliberate: the gap between the US regulatory frame the reader carries and the EU obligation arriving is exactly where the exposure is hardest to discover in advance. The same legal architecture catches EU-headquartered originators and EU generics manufacturers with Asian API exposure in identical terms, and in absolute volume the population of EU sponsors caught by Art. 22(4) of the Directive is the larger one; what differs is the commercial and contractual baseline they bring to the obligation, not the obligation itself.