1.

The Consiglio di Stato (Council of State, Italy) has referred to the Court of Justice several questions for a preliminary ruling concerning the interpretation of Article 101 TFEU in the context of a dispute the unusual characteristics of which may be summarised as follows.

2.

A certain undertaking has developed two medicines, one for the treatment of cancer, the other for the treatment of ophthalmological conditions. The medicines are based on different active substances that are nevertheless obtained from the same antibody and have the same therapeutic mechanism. The undertaking decided to market the cancer treatment itself and entrusted the marketing of the ophthalmological medicine to another undertaking by means of a licensing agreement.

3.

The marketing authorisation (‘MA’) for the cancer treatment was granted approximately two years earlier than the MA for the ophthalmological medicine. During the interval between the grant of each of the two MAs, a number of medical practitioners gave their patients the cancer treatment, in weaker doses, to treat ocular pathologies. This usage for therapeutic indications and in accordance with methods not covered by the Summary of Product Characteristics (‘SPC’) or, therefore, by the MA for the medicine — so-called ‘off-label use’ — continued even after the grant of the MA for the ophthalmological medicine because of the substantially lower treatment costs.

4.

The Autorità Garante della Concorrenza e del Mercato (AGCM) (the Italian competition authority) found that the two undertakings in question had conspired to communicate to the pharmaceutical regulatory authorities, to medical professionals and to the general public statements to the effect that the cancer treatment when used off label was less safe than the ophthalmological medicine. According to the AGCM, the undertakings had no scientific evidence to support these statements and had made them with the aim of discouraging the off-label use of the cancer treatment and consequently increasing sales of the ophthalmological medicine. The AGCM considered that this collusive conduct was a restriction of competition by object, within the meaning of Article 101(1) TFEU, and fined the undertakings accordingly.

5.

Having failed in their action at first instance to challenge that decision, the undertakings brought an appeal before the referring court. It is in that context that the referring court questions the Court of Justice about the interplay between the regulatory framework for the placing of medicinal products on the market and EU competition law. In particular, the Court is called upon to clarify to what extent and on what basis the legal uncertainty regarding the lawfulness of prescribing and marketing a medicine with a view to its off-label use and the scientific uncertainty surrounding the risks associated with such use come into play in the application of Article 101 TFEU.

6.

Regulation (EC) No 772/2004, ( 2 ) which was applicable at the time of the material facts in the main proceedings, provided for the block exemption of certain technology transfer agreements.

7.

Pursuant to Article 1(1)(j)(ii) of that regulation, ‘competing undertakings on the relevant product market’ are ‘undertakings which, in the absence of the technology transfer agreement, are both active on the relevant product and geographic market(s) on which the contract products are sold without infringing each other’s intellectual property rights (actual competitors on the product market) or would, on realistic grounds, undertake the necessary additional investments or other necessary switching costs so that they could timely enter, without infringing each other’s intellectual property rights, the(se) relevant product and geographic market(s) in response to a small and permanent increase in relative prices (potential competitors on the product market); the relevant product market comprises products which are regarded by the buyers as interchangeable with or substitutable for the contract products, by reason of the products’ characteristics, their prices and their intended use’.

8.

Regulation (EC) No 726/2004 ( 3 ) established a centralised procedure for the authorisation of medicinal products at EU level.

9.

In accordance with Article 3(1) of that regulation ‘no medicinal product appearing in the annex may be placed on the market within the [European Union] unless [an MA] has been granted by the [European Union] in accordance with the provisions of this regulation’. Point 1 of the annex, which concerns ‘medicinal products to be authorised by the [European Union]’, includes medicines developed by means of certain biotechnological processes.

10.

According to Article 13(1) of the regulation, an MA granted on completion of the centralised procedure is valid throughout the European Union and confers the same rights and obligations in each of the Member States as an MA granted by that Member State in accordance with Directive 2001/83/EC. ( 4 )

11.

In so far as concerns the content of an application for an MA, Article 6(1) of Regulation No 726/2004 refers to the particulars listed, inter alia, in Article 8(3) of Directive 2001/83. Point (j) of Article 8(3) in particular mentions the SPC. In accordance with Article 11 of the directive, the SPC must specify the dosage and the pharmaceutical form of the medicinal product, the qualitative and quantitative composition of all of its constituents, the therapeutic indications, the posology and method of administration, the contra-indications, special warnings and special precautions for use, the adverse reactions and the special precautions for storage and the duration thereof.

12.

Article 16(2) of Regulation No 726/2004, in the version which has applied since 2 July 2012, ( 5 ) provides that the holder of an MA must immediately provide the European Medicines Agency (EMA), the European Commission and the Member States with any new information which might entail the amendment of the particulars or documents referred to, inter alia, in Article 8(3) of Directive 2001/83. The information to be provided must ‘include both positive and negative results of clinical trials or other studies in all indications and populations, whether or not included in the [MA], as well as data on the use of the medicinal product where such use is outside the terms of the [MA]’.

13.

Regulation No 726/2004 also instituted a system of pharmacovigilance for the medicinal products authorised under it. As is clear from Article 24(1) of the regulation, in the version which has applied since 2 July 2012, that system serves to collate information ‘on suspected adverse reactions in human beings arising from use of the medicinal product within the terms of the [MA] as well as from uses outside the terms of the [MA]’.

14.

In particular, Article 21(1) of Regulation No 726/2004, in the version which has applied since 2 July 2012, provides that ‘the obligations of [MA] holders laid down in Article 104 of Directive 2001/83/EC shall apply to … holders [of MAs] for medicinal products for human use authorised in accordance with this regulation’.

15.

Article 104 of Directive 2001/83, as it results from an amendment which the Member States were to transpose by 21 July 2012, ( 6 ) is worded as follows:

‘1.   The [MA] holder shall operate a pharmacovigilance system for the fulfilment of his pharmacovigilance tasks equivalent to the relevant Member State’s pharmacovigilance system provided for under Article 101(1).

2.   The [MA] holder shall by means of the pharmacovigilance system referred to in paragraph 1 evaluate all information scientifically, consider options for risk minimisation and prevention and take appropriate measures as necessary. …

…’

16.

In accordance with Article 49(5) of Regulation No 726/2004, an MA holder may not communicate information relating to pharmacovigilance concerns to the general public without giving prior or simultaneous notification to the EMA. An MA holder must, in any event, ensure that such information ‘is presented objectively and is not misleading’.

17.

By decision of 27 February 2014 (‘the AGCM’s decision’), a copy of which was placed on the file submitted by the referring court to the Court of Justice, the AGCM found that the companies F. Hoffmann-La Roche Ltd (‘Roche’) and Novartis AG, through their subsidiaries Roche SpA (‘Roche Italia’) and Novartis Farma SpA (‘Novartis Italia’), had put in place, in breach of Article 101 TFEU, a horizontal agreement restricting competition. According to the AGCM’s findings, that agreement was designed to achieve an artificial differentiation of the medicinal products Avastin and Lucentis by manipulating the perception of the risks associated with the use of Avastin in the field of ophthalmology. The AGCM imposed on the four companies administrative fines amounting to approximately EUR 180 million in total.

18.

Roche, Roche Italia, Novartis and Novartis Italia (‘the applicants in the main proceedings’) challenged that decision before the Tribunale amministrativo regionale per il Lazio (Regional Administrative Court, Lazio, Italy) which joined their actions and dismissed them by judgment of 2 December 2014.

19.

The applicants in the main proceedings brought an appeal before the Consiglio di Stato (Council of State) in order to have the judgment set aside.

20.

In that context, the Consiglio di Stato (Council of State) states that the medicinal products Avastin and Lucentis were developed by Genentech Inc., a biotechnology company under the sole control of the Roche group, in the course of a single research programme. That programme was launched following the discovery of a protein produced by the human body (named ‘vascular endothelial growth factor’ (VEGF)) that is responsible for the formation of abnormal blood vessels which contribute to the growth of certain cancerous tumours.

21.

The researchers at Genentech discovered that inhibiting the action of VEGF, by means of an antibody, could be used in the treatment of certain types of cancer. They then succeeded in obtaining an anti-VEGF antibody that could be administered to humans, bevacizumab, which became the active substance in the medicine Avastin.

22.

The researchers also examined other diseases connected with the action of VEGF, including a common eye disease known as age-related macular degeneration (AMD). However, the researchers considered bevacizumab to be unsuitable, in terms of safety and efficacy, for the treatment of AMD and other ocular vascular pathologies.

23.

Genentech therefore decided to develop an anti-VEGF treatment specifically for the ophthalmological field. The researchers eventually identified an anti-VEGF antibody fragment, named ranibizumab, which became the active substance in the medicine Lucentis. Ranibizumab is eliminated by the body faster than bevacizumab and it has smaller dimensions, which facilitate its penetration of the retina and its capacity to bind itself to the VEGF.

24.

Genentech, which is commercially active only in the United States, granted a licence to exploit Avastin to its parent company Roche and, since Roche is not active in the field of ophthalmology, granted a licence to exploit Lucentis to the Novartis group, so that the two companies could register and market the two medicinal products in the rest of the world. The licensing agreement for Lucentis was concluded in June 2003.

25.

The medicinal products Avastin and Lucentis were granted MAs by the EMA for the treatment of certain tumorous diseases and certain ophthalmic diseases respectively.

26.

On 26 September 2005, the Agenzia Italiana del Farmaco (AIFA) (the Italian Medicines Agency) transposed the MA granted at EU level for Avastin and decided that the cost of the medicine could be reimbursed by the Servizio Sanitario Nazionale (SSN) (the Italian National Health Service).

27.

An MA for Lucentis, for the treatment of AMD, was not obtained from AIFA until 31 May 2007. Initially non-reimbursable because of the lack of any agreement between AIFA and Novartis regarding the reimbursement price, Lucentis was approved for reimbursement by the SSN on 4 December 2008.

28.

During the period between the market launch of Avastin and that of Lucentis, certain medical practitioners, following early administrations of Avastin for the treatment of cancer, observed that the health of patients suffering from both a cancerous tumour and AMD improved also in so far as the latter disease was concerned.

29.

Since Avastin was, during that period, the only anti-VEGF medicine available on the market, some medical practitioners administered it intravitreally (by injection into the eye) to patients suffering from AMD, even though, under the terms of its SPC, Avastin was not authorised for that therapeutic indication or for that method of administration. This off-label use of Avastin spread worldwide. It continued after Lucentis had been placed on the market because of the lower cost of therapies using Avastin.

30.

Italian legislation permits, under certain circumstances, the reimbursement of medicines used off label. At the time of the facts material to the AGCM’s decision, such reimbursement was subject to the fulfilment of two conditions. First, no viable alternative therapy must have been authorised for the treatment of the pathology in question. Second, the AIFA must have included the off-label use in question in the list of medicinal products covered by the SSN, the so-called ‘List 648’. ( 7 )

31.

Following the grant of the MA for Avastin and after its off-label use in ophthalmology had become widespread in Italy, in May 2007, the AIFA included that use in List 648 in connection with the treatment of exudative macular diseases (AMD, retinal vein occlusion (RVO), diabetic macular edema (DME), myopic macular degeneration (MMD)) and neovascular glaucoma, there being no valid therapeutic alternative at that time for treating those diseases.

32.

Subsequently, following the approval and authorisation for reimbursement in Italy of the medicinal products Lucentis and Macugen for the treatment of AMD (at the end of 2008) and then the approval and authorisation for reimbursement in Italy of the medicinal product Ozurdex, for the treatment of RVO (in July 2011), AIFA decided that there would be no reimbursement for the use of Avastin in the treatment of those diseases. Finally, on 18 October 2012, AIFA removed Avastin from List 648 entirely, referring for that purpose to certain amendments which the EMA had made to the SPC for that medicine on 30 August 2012. As is apparent from the case file submitted to the Court of Justice, those amendments concerned the addition of certain special warnings and precautions for use relating to the intravitreal use of Avastin.

33.

The AGCM has emphasised that, the cost of Avastin having long been covered by the SSN in connection with various ophthalmological uses, the medicine became, at least during the period between its inclusion on List 648 and the commencement of the AGCM’s procedure, the principal anti-VEGF medicine used in Italy in the treatment of ocular vascular pathologies, in terms of the number of patients treated. Because of that extremely widespread off-label use of Avastin, it became the principal competitor of Lucentis.

34.

According to the AGCM, the applicants in the main proceedings had put in place ‘a single and complex horizontal agreement implemented through a multitude of concerted practices’. The purpose of that agreement was to achieve an ‘artificial differentiation’ between the medicines Avastin and Lucentis — which are ‘equivalent medicinal products in all respects in the field of ophthalmology’ — by manipulating the perception of the risks involved in the use of Avastin in ophthalmology so as to influence demand in favour of Lucentis. The agreement had been implemented by ‘the production and dissemination of opinions which could give rise to public concern regarding the safety of the intravitreal use of Avastin’, while at the same time ‘downplaying the value of scientific evidence to the contrary’.

35.

The AGCM found that the companies had exaggerated the risks involved in the intravitreal use of Avastin and had at the same time alleged that Lucentis had a better safety profile than Avastin. The companies had also relied on the fact that only Lucentis had an MA for ophthalmological indications and that no application had yet been made for an MA for Avastin for such indications.

36.

The applicants in the main proceedings had thus sought to ‘prevent the off-label use of Avastin from eroding the on-label use of Lucentis’, which was ‘the more costly product … the sale of which generates profits for both companies’. The agreement also included ‘the common interest of the Roche and Novartis groups in the amendment of the [SPC] for Avastin, currently before the EMA, and in the subsequent issue of a formal communication to medical professionals (a ‘direct healthcare professional communication’ or DHPC) — one that had been initiated by Roche, as [MA] holder for Avastin … — that would have a direct effect on the concerted plan to create artificial differentiation’.

37.

Still according to the AGCM, the agreement in question was intended to maximise unlawfully the respective revenues of the Roche and Novartis groups which came, for the Novartis group, from the direct sale of Lucentis and from its 33% shareholding in Roche and, for the Roche group, from royalties received on such sales through the intermediary of its subsidiary Genentech.

38.

The concerted practices identified by the AGCM, ‘although dependent on the existence of vertical licensing relationships, had taken place outside those relationships’.

39.

The AGCM considered that the arrangement constituted an unlawful market-sharing agreement and therefore constituted a restriction of competition by object, within the meaning of Article 101(1) TFEU. It had been ‘put into effect, having an impact on the therapeutic decisions taken by medical practitioners and on the resulting purchasing policies for the medicines Avastin and Lucentis’. The agreement had ‘caused an immediate slowdown in the growth of Avastin and a consequent shift in demand toward the more expensive Lucentis, which [had] resulted in a cost increase for the SSN of nearly EUR 45 million in 2012 alone’.

40.

The AGCM consequently considered that ‘the infringement complained of … must be regarded as very serious’, in particular, because of its unlawful object, because it had ‘been implemented’ and had ‘had definite effects on the economic equilibrium of the health system as a whole’, and also because of the fact that the combined market share of the applicants in the main proceedings of the Italian market for medicinal products for the treatment of ocular vascular pathologies was over 90%.

41.

The concerted practices identified by the AGCM had begun in June 2011 at the latest, when Roche initiated the formal procedure before the EMA to obtain the amendment of the SPC for Avastin and the consequent issue of a formal communication. The concerted practices had not ceased at the time when the AGCM adopted its decision.

42.

In that context, the Consiglio di Stato (Council of State) decided to stay the proceedings and to refer the following questions to the Court of Justice for a preliminary ruling:

43.

Written observations have been submitted to the Court by Roche, Roche Italia, Novartis and Novartis Italia, the Associazione Italiana delle Unità Dedicate Autonome Private di Day Surgery e dei Centri di Chirurgia Ambulatoriale (Aiudapds), the Società Oftalmologica Italiana (SOI) — Associazione Medici Oculisti Italiani (AMOI) (‘SOI-AMOI’), Altroconsumo, the Coordinamento delle associazioni per la tutela dell’ambiente e dei diritti degli utenti e consumatori (Condacons), the AGCM, the Regione Emilia-Romagna (the Region of Emilia-Romagna, Italy), the Italian, Irish and French Governments and the Commission.

44.

Roche, Roche Italia, Novartis, AIUDAPDS, SOI‑AMOI, Altroconsumo, the AGCM, the Region of Emilia-Romagna, the Italian Government and the Commission attended the hearing on 3 May 2017.

45.

What lies behind the present case is a situation marked by the development of a wide-scale medical practice of prescribing a certain medicine for off-label use. ( 8 ) This practice developed, against the wishes of the holder of the MA for that medicine, at the initiative of those who create the demand for it, that is to say prescribing medical practitioners, who were further encouraged by the authority’s decision to allow reimbursement by the SSN of the medicinal product in question.

46.

The prescribing of Avastin for the treatment of ocular vascular pathologies was initially intended to fill a therapeutic lacuna, there being no other equally efficacious medicinal products having an MA for such therapeutic indications. The practice nevertheless continued after the placing on the market and the approval for reimbursement of such medicines. That was essentially for economic reasons, given the considerable price difference between therapies based on Avastin and those based on Lucentis. According to the AGCM, the difference in price per millilitre of the two medicinal products meant that an intravitreal injection of Avastin would have cost at least ten times less than an injection of Lucentis.

47.

The off-label use of medicinal products is a medical practice which varies in extent depending on the therapeutic field and the Member State concerned. ( 9 ) EU law acknowledges this reality and lays down certain provisions, upstream and downstream of off-label use, which restrict the possibilities for placing medicinal products intended for such use on the market ( 10 ) and impose on MA holders certain pharmacovigilance obligations in relation to off-label use. ( 11 )

48.

On the other hand, EU law does not govern the prescribing of medicinal products for off-label use. ( 12 ) That practice falls within the scope of the therapeutic freedom of medical practitioners, subject to any restrictions imposed on that freedom by the Member States in the exercise of their power to define their health policies. ( 13 ) Equally, the decision to approve a medicine used off label for reimbursement by the social security systems lies, in principle with the Member States. ( 14 )

49.

In this context, the Member States have adopted diverging policies on the regulation of off-label uses of medicines in general and of Avastin in particular. Some have chosen to authorise the reimbursement of certain medicines prescribed off label or even to stipulate temporary recommendations for their use. ( 15 ) Significant judicial proceedings have arisen regarding the legality, with regard to EU law, of such domestic regulations. ( 16 ) Moreover, the Consiglio di Stato (Council of State) has, in another case pending before it, referred to the Court of Justice a question for a preliminary ruling concerning the compatibility with EU law of national measures which, for economic reasons, provide for the reimbursement of medicines prescribed off label, such as Avastin. ( 17 )

50.

Some, like the applicants in the main proceedings, argue in substance that national policies which authorise, or even encourage, the off-label prescribing of medicines for budgetary reasons run counter to the logic which underlies the EU regulatory framework for the placing of medicinal products on the market. ( 18 ) Since only the uses stipulated in the MA will have been the subject of the preclinical tests and clinical trials necessary for obtaining the MA, ( 19 ) uses which have not been validated by such trials should, at least, remain the exception.

51.

Others, such as the Region of Emilia-Romagna and the Irish Government in the present case, consider that the off-label use of a medicinal product for certain therapeutic indications is necessary where, despite evidence of the efficacy and safety of such use, the holder of the MA for the medicine does not take the necessary steps to extend the MA to cover it. The Region of Emilia-Romagna, the AGCM, SOI-AMOI and the Italian Government claim that off-label use is necessary, sometimes even when another medicine exists the MA for which covers the indications in question, in order to ensure access to care and to avoid excessively burdening the budgets of social security systems.

52.

It is not for me, in the context of the present case, to take a position in this debate or to express my thoughts on the merits of the policies of the Member States regarding the regulation of the off-label use of medicinal products. I shall therefore confine myself to examining whether, and if so to what extent, Article 101 TFEU protects the market dynamics which have resulted from off-label use.

53.

In this connection, it would seem useful to me to begin by outlining the main issues raised by the five questions referred by the national court.

54.

To begin with, the purpose of the second, third and fourth questions is to enable the national court to determine whether the legal barriers which arise from the provisions governing the placing on the market of medicinal products for off-label use preclude the substitutability of Avastin and Lucentis for the treatment of eye diseases and, consequently, their belonging to the same product market.

55.

In its reasoning relating to these questions, Roche argues that, having regard to those legal barriers, such products are not part of the same market and, more generally, are not in competition with each other. Accordingly, there could be no question of a restriction on competition resulting from the collusive conduct identified by the AGCM (‘the collusive conduct at issue’).

56.

Next, by its first question, the national court seeks to establish whether Genentech and Novartis must be regarded as competing undertakings in the context of the licensing agreement relating to Lucentis. If they are not, the national court questions the Court of Justice about the relevance, for the purposes of the application of Article 101 TFEU, of the fact that the collusive conduct at issue arose in the context of a licensing agreement between undertaking that are not in competition.

57.

The line of argument put forward by the applicants in the main proceedings highlights the issues raised by this question. According to them, the licensing agreement relating to Lucentis binds undertakings that are not in competition. The restrictions on the off-label use of Avastin pursued by means of the collusive conduct at issue (‘the restrictions at issue in the main proceedings’), while not expressly provided for in the licensing agreement, arose in the course of the agreement’s implementation. They allege that, in the case of licensing agreements between non-competing undertakings, the elimination of competition between licensor and licensee falls outside the scope of application of Article 101(1) TFEU or is, at least, exempted under Article 101(3) TFEU.

58.

Lastly, the fifth question invites the Court to clarify whether the collusive conduct at issue may, in any event, be classified as a restriction of competition by object, even though the scientific debate concerning the comparative safety and efficacy of Avastin and Lucentis in the field of ophthalmology had not been decided at the material time.

59.

It is in that order that I propose to analyse the questions referred for a preliminary ruling, after dismissing the main objections of inadmissibility that have been raised against them.

60.

The AGCM, Aiudapds and the Region of Emilia-Romagna dispute the admissibility of the questions referred, essentially on the ground that the statement of facts and law set out in the order for reference is deficient and incomplete, merely describing the arguments advanced by the applicants in the main proceedings, which are contested by other interested parties, and omitting important factual elements.

61.

The AGCM emphasises in particular that the order for reference fails to mention that Avastin has been used in the field of ophthalmology worldwide since 2005, without any adverse events of statistical relevance having been notified, so much so that the World Health Organisation (WHO) considers bevacizumab (the active substance in Avastin) as the only anti-VEGF medicine essential in ophthalmology. ( 20 ) The order for reference also says nothing about the fact that, in 2014, AIFA reinstated Avastin on List 648 for the treatment of ocular vascular pathologies.

62.

According to the AGCM and the Region of Emilia-Romagna, the Court of Justice is not in a position to give a useful answer to the referring court. The AGCM also submits, as does Aiudapds, that, given the incomplete and partly incorrect presentation of the facts, the questions referred are hypothetical.

63.

I would recall in this connection that, according to settled case-law, the Court may refuse to rule on a question submitted by a national court only where it is quite obvious that the interpretation of EU law that is sought bears no relation to the actual facts of the main action or its subject matter, where the problem is hypothetical, or where the Court does not have before it the factual or legal material necessary to give a useful answer to the questions submitted to it. ( 21 )

64.

As for the last of those grounds for refusing to rule on a request for a preliminary ruling, the Court has stated that the information provided to it in an order for reference serves not only to enable it to provide answers which will be of use to the national court, but also to enable the governments of the Member States and other interested parties to submit observations in accordance with Article 23 of the Statute of the Court of Justice of the European Union. To that end, the national court must define the factual and legislative context of the questions which it is asking or, at the very least, explain the factual assumptions on which those questions are based. ( 22 )

65.

As regards, first of all, the arguments alleging the incomplete nature of the statement of facts and law set out by the national court, the Court considers that, even where there are gaps in the order for reference, it does have sufficient information to reply usefully to the questions referred if the order for reference enables it to determine the scope of the questions referred. ( 23 ) The order for reference in the present case satisfies that condition in my view. The Court is, therefore, in a position to provide answers which will be of use to the referring court and the interested parties have been able to submit their observations to the Court, as is clear from the pleadings that have been lodged. ( 24 )

66.

Secondly, as regards the arguments alleging the incorrect nature of the description of the material facts, it is not for the Court of Justice but for the national court to ascertain the facts which have given rise to the dispute. ( 25 ) It is not for the Court of Justice to verify the accuracy of the legal and factual background which the national court is responsible for defining. ( 26 ) It is, in principle, required to base its considerations on the premisses which the national court regards as having been established. ( 27 )

67.

Consequently, I consider that the questions referred for a preliminary ruling are admissible.

68.

The second, third and fourth questions concern the extent to which the regulatory framework for the placing of medicinal products on the market must be taken into account in the definition of the relevant product market. By its second and third questions, which I propose to examine together, the referring court asks, in substance, whether, in the pharmaceutical sector, the definition of the relevant market is necessarily circumscribed by the content of MAs. By its fourth question, the referring court asks the Court of Justice about the relevance, in this context, of uncertainty regarding the lawfulness of marketing medicinal products which have been repackaged with a view to their off-label use.

69.

In the present case, the AGCM has defined the relevant product market as including all medicinal products for the treatment of ocular vascular pathologies. ( 28 ) That definition is not called into question in the present case. The only point under discussion is whether Avastin belongs to that market or not.

70.

It is clear from both the legislation ( 29 ) and the case-law ( 30 ) that a relevant product market comprises all those products which are regarded as interchangeable or substitutable by the consumer, by reason of their characteristics, their prices and their intended use. ( 31 )

71.

According to the case-law, it is also necessary, when making such an assessment, to take into account not only the objective characteristics of the products, by virtue of which they are particularly suitable for satisfying the constant needs of consumers, but also the conditions of competition and the structure of supply and demand. ( 32 )

72.

In accordance with those principles, the definition of the relevant product market depends not on criteria established in advance by the legal rules governing the conduct of economic actors, but on the objective characteristics of the products and on the actual competitive conditions surrounding such conduct. Those conditions will include legal rules to the extent that they are capable of affecting the degree of interchangeability of the goods in question, but are not limited to such legal rules. Other circumstances can, as the case may be, indicate the existence of effective competitive constraints.

73.

In this case, the legal framework governing the placing on the market — and the prescribing ( 33 ) — of medicinal products may entail certain legislative obstacles to the substitutability of a medicine used off label for a medicine used on label for the same therapeutic indications. ( 34 ) Nevertheless, such obstacles are not insurmountable, nor therefore are they necessarily decisive in the definition of the relevant market.

74.

In light of those considerations, I think that, where the competitive conditions actually observed indicate that there is effective demand-side substitutability between a medicinal product used off label for certain therapeutic indications and another medicinal product that has received an MA in respect of the same therapeutic indications, the two products belong to the same product market (section 1 below). That applies even where the lawfulness of prescribing and marketing the first medicinal product with a view to off-label use is uncertain (section 2 below).

75.

As has been argued by all of the interested parties, with the exception of the applicants in the main proceedings, the fact that the MA for a medicinal product does not cover certain therapeutic indications does not mean that that medicine cannot have a degree of interchangeability with medicines which are authorised for those indications that is sufficient to create an effective competitive constraint on them.

76.

Certainly, the content of an MA will, in principle, influence the substitutability of different medicinal products for the same therapeutic use. In so far as prescription medicines are concerned, demand is generally determined not by the preferences of final consumers (that is to say, patients) but by the decisions of medical practitioners, and an MA is likely, at the very least, to guide medical practitioners in their choice of appropriate treatment for their patients. That is especially true where national law limits the possibilities for prescribing medicinal products off label or for obtaining reimbursement and lays down specific rules governing medical liability in the event that harm is caused by the off-label use of a medicinal product.

77.

However, the practice of medical practitioners in prescribing medicines, coupled, as the case may be, with administrative decisions concerning the approval for reimbursement of medicines prescribed off label, may be at the origin of competitive dynamics which demonstrate the actual interchangeability of two medicinal products independently of the content of their respective MAs. Indeed, if the content of an MA is limited by the application that was made by its holder to the pharmaceutical regulatory authorities, ( 35 ) that application will not necessarily cover all the possible uses to which medical practitioners may put the medicine in question in the exercise of their therapeutic freedom. ( 36 )

78.

In the present case, it is apparent from the order for reference that Avastin was, at the time of the collusive conduct at issue, very often prescribed for ophthalmological indications. Moreover, when the concerted practices identified by the AGCM began (that is to say, in June 2011), Avastin was still included in the list of medicinal products which the SSN reimbursed for the treatment of neovascular glaucoma and all exudative macular diseases with the exception of AMD. ( 37 )

79.

Those circumstances reveal that Avastin, when used off label, exerted an effective competitive constraint on Lucentis. In accordance with the principles mentioned in points 70 and 71 of this Opinion, that constraint must be taken into consideration in the definition of the relevant product market.

80.

This approach reflects the approach that the Commission adopted in certain merger control decisions in which it took into account medicinal products used off label in its analysis of the actual competitive dynamics for the purposes of defining the relevant product market. ( 38 )

81.

Moreover, if the definition of the relevant product market were systematically limited by the content of MAs, pharmaceutical companies would, in practice, have carte blanche to reach agreements, prior to the placing of their medicines on the market, to share markets by ensuring that there was no overlapping of the therapeutic indications covered by their respective MA applications, as indeed Aiudapds, SOI-AMOI, Altroconsumo, Condacons and the Italian Government have argued. The market would then be defined with no account being taken of the interchangeability of medicinal products on the demand side, in breach of the principles set out in points 70 and 71 of this Opinion.

82.

The words ‘establish … whether’ used in the fourth question reflect the uncertainty surrounding the lawfulness of marketing Avastin once it has been repackaged with a view to its use in ophthalmology. That issue has been hotly debated in the written and oral submissions of the interested parties. According to the applicants in the main proceedings, that activity is unlawful in numerous cases, or even most of the time. Other interested parties, such as the AGCM, SOI-AMOI, the Region of Emilia-Romagna and the Italian Government, dispute that allegation. ( 39 )

83.

The applicants in the mains proceedings have also raised the question of the impact on the definition of the relevant market of the alleged infringement of the provisions of Italian law which limit the freedom of medical practitioners to prescribe off-label medicines.

84.

Under Italian law, that practice is, they allege, permitted only where, on the basis of an individual assessment, there is no authorised medicinal product with which a given patient can be treated effectively. ( 40 ) A medicinal product prescribed off label would therefore not be interchangeable with a medicinal product prescribed on label for the same indications, but subsidiary to it. The AGCM, SOI-AMOI, Condacons, the Region of Emilia-Romagna and the Italian Government do not share that view and suggest a different interpretation of Italian law. ( 41 )

85.

Since this particular question could also be relevant to the resolution of the dispute in the main proceedings, ( 42 ) I shall read the fourth question referred for a preliminary ruling as asking whether or not there is any need, when defining the relevant product market, to verify not only whether the marketing of medicinal products for off-label use, but also whether prescribing them for off-label use is in accordance with the applicable legal framework.

86.

In my view, the principles recalled in points 70 and 71 above imply that uncertainty regarding the lawfulness of prescribing or marketing medicinal products with a view to their off-label use for certain therapeutic indications does not, in itself, mean that such medicines do not form part of the same market as medicines authorised for those indications.

87.

Of course, the competition authorities and the courts responsible for applying the competition rules should take account of any such uncertainty where it is capable of precluding the interchangeability of such medicinal products. Nevertheless, if they find that a medicinal product is in fact widely used off label in spite of such uncertainty, they may validly find that such a medicine is interchangeable with medicines used on label for the same indications and therefore belongs to the same product market as them.

88.

They need not, in order to justify such a conclusion, remove that uncertainty by themselves assessing the lawfulness of prescribing and marketing medicinal products used off label. Indeed, such an exercise is unrelated to the application of the competition rules and does not normally fall within the competence of the authorities entrusted with their application. ( 43 ) As the AGCM, SOI-AMOI, the Italian Government and the Commission have emphasised, EU competition law pursues independent objectives distinct from those which pharmaceutical legislation seeks to achieve.

89.

The approach I advocate is also consistent with that taken in the judgment in Slovenská sporiteľňa, ( 44 ) in which the Court offered certain points of clarification concerning the application of Article 101 TFEU to a cartel between undertakings the purpose of which was to remove from the market in question a third undertaking whose activity on that market was alleged to be unlawful. Without first verifying whether the services offered by the eliminated undertaking belonged to the same market as those offered by the undertakings party to the cartel, the Court held that the fact that the eliminated undertaking was allegedly operating unlawfully on the market in question at the time when the cartel was concluded was irrelevant to the application of Article 101(1) TFEU. To support that conclusion, the Court emphasised that it was for the public authorities, and not private undertakings, to ensure compliance with statutory provisions, the application of which could call for complex assessments not within the area of responsibility of private undertakings. ( 45 ) The Court then considered whether that fact could justify the grant of an exemption under Article 101(3) TFEU. ( 46 )

90.

That reasoning presupposes that the alleged unlawfulness of the offer of certain goods or services does not, in itself, prevent them from belonging to the same market as other goods or services the lawfulness of whose supply is not in question. ( 47 )

91.

By its first question, the referring court asks whether the parties to a licensing agreement are to be regarded as competing undertakings when the licensee operates on the relevant market only by reason of that agreement. If that question is answered in the negative, the referring court asks the Court of Justice, essentially, about the consequences, for the analysis with regard to Article 101(1) and (3) TFEU of collusive conduct such as that at issue in the main proceedings, of the fact that the conduct occurred in the context of a contractual licensing relationship between undertakings that are not in competition.

92.

An agreement for the grant of a licence concerning intellectual property rights, such as the agreement between Genentech and Novartis relating to Lucentis, is, in principle, a ‘technology transfer agreement’ within the meaning of Article 1(1)(b) of Regulation No 772/2004. ( 48 )

93.

As is clear from Article 1(1)(j)(ii) of that regulation, the parties to a technology transfer agreement are regarded as competing undertakings in the market in which the goods manufactured with the aid of the technology transferred under the licence (referred to as ‘contract products’ ( 49 )) are sold if, absent the agreement, they would have been actual or potential competitors in that market.

94.

Consequently, the parties to a licensing agreement are not regarded as competing undertakings when the licensee operates on the relevant market only by virtue of the agreement, in the absence of which it would have been neither an actual competitor nor a potential competitor of the licensor.

95.

In the present case, no party disputes that, in the absence of the licensing agreement relating to Lucentis, Novartis would not have been an actual or potential competitor of Genentech in the market for medicinal products for the treatment of ocular vascular pathologies. Indeed, there is nothing in the file submitted to the Court to show that Novartis would even have commenced research and development with a view to creating a medicinal product for the treatment of such pathologies.

96.

Thus, the collusive conduct at issue occurred in the context of a contractual licensing relationship between non-competing undertakings, in the absence of which there would have been no reason for it, as Roche emphasises.

97.

That said, for reasons which I shall set out below, the collusive conduct at issue cannot be excluded from the prohibition laid down in Article 101(1) TFEU or benefit from exemption under Article 101(3) TFEU on the ground that the restrictions at issue in the main proceedings are allegedly similar to restrictions of competition that is exerted by a licensor with regard to a licensee set out in a licensing agreement between non-competing undertakings.

98.

As is clear from their wording, the second and third parts of the first question concern the application of Article 101(1) and (3) TFEU to ‘restrictions of competition between … licensor and … licensee’. In order to provide the national court with a useful answer, it seems to me necessary to clarify somewhat the nature and scope of the restrictions at issue in the main proceedings, to which this question refers, in the light of the factual context described in the order for reference.

99.

I would emphasise, in the first place, that the restrictions at issue were, more specifically, restrictions of the competition that is exerted with regard to a licensee by means of the demand for and use by third parties, in a form and for purposes not contemplated by the licensor, of a product initially manufactured and marketed by the licensor. ( 50 )

100.

In the second place, the interested parties disagree as to whether those restrictions related to competition between two products which rely on the same technology (‘intra-technology competition’) or between two products which rely on different technology (‘inter-technology competition’).

101.

The significance of that distinction is that certain restrictions of intra-technology competition, in so far as they are deemed necessary for the dissemination of a new technology and, consequently, for the strengthening of inter-technology competition, fall outside the scope of Article 101(1) TFEU. ( 51 )

102.

Altroconsumo maintains that Avastin and Lucentis do not rely on the same technologies and that the collusive conduct at issue accordingly restricted inter-technology competition between those two products. The facts to which the referring court draws the attention of the Court of Justice do not permit the accuracy of that allegation to be verified. Roche disputes the allegation and argued at the hearing that Avastin and Lucentis are manufactured on the basis of the same patents, which thus cover both the anti-VEGF medicines developed by Genentech.

103.

Subject to verification by the referring court, I shall assume that the two medicinal products were both produced on the basis of rights to the technology that was licensed under the licensing agreement relating to Lucentis, and indeed the answers which I shall suggest will be all the more valid if it proves that the medicines do not in fact rely on the same technology. ( 52 )

104.

In so far as concerns the second part of the first question, I consider that, even if they had been set out expressly in the licensing agreement relating to Lucentis, the restrictions at issue in the main proceedings could not escape the prohibition laid down in Article 101(1) TFEU on the ground that, as the applicants in the main proceedings contend, they restricted competition exerted by a licensor with regard to a licensee.

105.

The doubts which the referring court has in this regard reflect a certain line of case-law according to which, where the conclusion or implementation of an agreement which is in itself pro-competitive, or at least neutral as regards competition, requires the insertion into that agreement of certain restrictions on the commercial independence of the parties, those restrictions do not fall within the scope of Article 101(1) TFEU. While the Court has not always used this terminology, this case-law upholds the doctrine of ‘ancillary restrictions’.

106.

This doctrine may be seen as drawing its origins from the judgment in LTM, ( 53 ) in which the Court emphasised the necessity, when considering the lawfulness of collusion, of analysing the situation which would have existed in the absence of the collusion. It held, with regard to the grant by a manufacturer to a distributor of an exclusive retailing right in a given territory, that ‘it may be doubted whether there is an interference with competition if the … agreement [in question] seems really necessary for the penetration of a new area by an undertaking’. The Court subsequently applied and developed that doctrine in a series of judgments, ( 54 ) including Nungesser and Eisele v Commission ( 55 ) and, most recently, MasterCard and Others v Commission. ( 56 )

107.

Indeed, the applicants in the main proceedings refer to paragraph 57 of the judgment in Nungesser and Eisele v Commission ( 57 ) in support of their argument that Article 101(1) TFEU does not apply to the collusive conduct at issue. In that judgment, the Court examined a so-called ‘open’ territorial exclusivity clause under which a licensor undertook not to grant other licences in respect of the territory granted and not to compete itself with the licensee in exploiting the rights relating to the licensed technology. According to the Court, the clause was necessary for the very existence of the licensing agreement since, without it, the licensee might have been deterred from accepting the risks associated with the exploitation of the licensed technology. In substance, the Court thus considered that, in order to promote the inter-technology competition which results from the dissemination of a new technology by means of a licensing agreement, ( 58 ) certain restrictions of intra-technology competition between undertakings capable of exploiting that technology may prove necessary. ( 59 )

108.

In that judgment, the Court also examined a so-called ‘closed’ exclusivity clause by which the parties to the licensing agreement proposed to eliminate all competition from third parties, such as parallel importers and licensees in other territories. The parties to the agreement were criticised for having initiated proceedings and exerted pressure on parallel importers in pursuance of that clause. The Court did not hold that that clause was necessary for the dissemination of a new technology. The closed exclusivity clause did not, therefore, escape the application of Article 101(1) TFEU ( 60 ). Nor could it be exempted under Article 101(3) TFEU, because it manifestly went beyond what was indispensable for achieving the gains in efficiency. ( 61 )

109.

According to Roche and Roche Italia, the restrictions at issue in the main proceedings are comparable to an exclusive licence under which the licensor undertakes not to compete with the licensee by producing under the technology rights transferred or selling products which rely on that technology. Consequently, the approach adopted in paragraph 57 of the judgment in Nungesser and Eisele v Commission ( 62 ) can be applied in the present case.

110.

I do not share that view.

111.

Indeed, as is clear from the observations made by the AGCM which are set out in the order for reference, and as the Italian Government and the Commission have emphasised, the purpose of the collusive conduct at issue was not to restrict the production or sale by Genentech or other companies in the Roche group of products which incorporate the technology licensed to Novartis. On the contrary, its purpose was to influence the actions of parties unconnected with the licensing agreement relating to Lucentis, that is to say, the pharmaceutical regulatory authorities and medical practitioners, so as to limit the use of Avastin in ophthalmology. In other words, the applicants in the main proceedings intended not to alter the supply of Avastin, but to alter the demand from medical practitioners (on whose judgment patients rely) who were prescribing Avastin off label. Indeed, it was as a result of that demand that Avastin entered into competition with Lucentis.

112.

Inasmuch as its purpose was to impede competitive dynamics independent of the wishes of the licensor and arising from sources not within the licensor’s control, ( 63 ) the collusive conduct at issue raises different issues from those raised by open exclusive licenses such as that examined by the Court in paragraph 57 of its judgment in Nungesser and Eisele v Commission. ( 64 )

113.

In my view, the restrictions at issue in the main proceedings should instead be treated similarly to the closed exclusive licence at issue in that judgment ( 65 ) in so far as Article 101(1) TFEU is concerned. Admittedly, what underpinned the Court’s approach was the objective of achieving the integration of geographic markets, which is not relevant in the present case. ( 66 ) However, I would observe that EU competition law focuses on combating the partitioning not only of the geographic markets, but also of the product markets on which undertakings are active. ( 67 ) At the very least, it cannot be inferred from that judgment that the elimination of all competitive pressure connected with products incorporating the licensed technology, including pressure from independent sources not under the control of the licensor, is ancillary to the conclusion or implementation of a licensing agreement.

114.

The conclusion which I advocate also follows from an examination of the restrictions at issue in the main proceedings in the light of the most recent case-law, which is the judgment in MasterCard and Others v Commission, ( 68 ) in which the Court summarised and refined the theory of ancillary restrictions.

115.

The Court first of all pointed out that, ‘if a given operation or activity is not covered by the prohibition rule laid down in Article [101(1) TFEU], owing to its neutrality or positive effect in terms of competition, a restriction of the commercial autonomy of one or more of the participants in that operation or activity is not covered by that prohibition rule either if that restriction is objectively necessary to the implementation of that operation or that activity and proportionate to the objectives of one or the other’. ( 69 )

116.

Next, the Court stated that the condition of objective necessity was fulfilled only where it was not possible to dissociate the restriction at issue from the main operation without jeopardising its existence and aims. That was the case where it would be impossible to carry out the operation in the absence of the restriction. On the other hand, the fact that the operation would, in the absence of the restriction, simply be more difficult to implement, or less profitable even, did not confer on the restriction the objective necessity required for it be classified as ancillary. ( 70 )

117.

Thus, in that judgment, a narrow interpretation was given of the doctrine of ancillary restrictions: without undermining the effectiveness of the prohibition laid down in Article 101(1) TFEU it could only apply to restrictions that are ‘strictly indispensable to the implementation of the main operation’. ( 71 )

118.

I doubt that restrictions such as those at issue in the main proceedings — even if they had been inserted into the licensing agreement — would constitute ancillary restrictions according to that case-law.

119.

First of all, they are not restrictions ‘of the commercial autonomy of one or more of the participants’ in a main operation, within the meaning of the judgment in MasterCard and Others v Commission. ( 72 ) Indeed, the restrictions which the Court classified as ancillary in that judgment and in its earlier case-law were invariably restrictions on the conduct of the parties to the principal operation themselves. ( 73 )

120.

According to the AGCM, the collusive conduct at issue, which admittedly involved the adoption by Roche and Roche Italia of a certain course of action in relation to the communication concerning the off-label use of Avastin, was aimed not at restricting the commercial independence of the parties to the licensing agreement relating to Lucentis, but to impede the competitive dynamics resulting from the actions of parties unconnected with that agreement. ( 74 )

121.

In the second place, I am not convinced that restrictions such as those at issue in the main proceedings are ‘objectively necessary to the implementation’ of a licensing agreement, again within the meaning of the judgment in MasterCard and Others v Commission. ( 75 )

122.

It seems to me difficult to contend that it would be impossible to conclude a licensing agreement granting technology rights for the production and/or marketing of a medicinal product that is authorised for certain therapeutic indications without an undertaking from the licensor to impede the competition that arises from the demand expressed by medical practitioners for a different medicinal product which incorporates the same technology but is prescribed off label for such indications. The fact that, in some cases, the demand for a medicine used off label may alter the demand for a medicine covered by a licensing agreement and thus render the exploitation of the rights to the licensed technology less profitable is not sufficient to establish that such a restriction is objectively necessary. ( 76 )

123.

That is all the more true where, as in the present case, the restrictions were agreed upon not in the licensing agreement, but under a concerted practice post-dating the conclusion of the licensing agreement by several years. In my view, that last fact serves to indicate that the restrictions at issue were not objectively necessary for the implementation of the licensing agreement. Moreover, where a licensee has already made the necessary investments for launching the contract products — such as those required in order to obtain an MA — I fail to see how the implementation of the agreement could not be continued without such restrictions.

124.

Should it be the case that Avastin and Lucentis do not incorporate the same technology, the collusive conduct at issue, a fortiori, will likewise not escape the application of Article 101(1) TFEU. ( 77 ) While restrictions relating to the exploitation of licensed technology by a licensor may fall outside the scope of that provision if they are objectively necessary for the conclusion of a licensing agreement, ( 78 ) that reasoning does not in any event apply to restrictions on the exploitation by a licensor of a different technology. On the contrary, the weakening of the competition arising from that other technology could erase the beneficial effect for competition which results from the dissemination of the new technology by means of a licensing agreement.

125.

In my opinion, the nature of the restrictions at issue in the main proceedings, and the fact that they occurred in the context of a licensing agreement between non-competing undertakings, as such, is no more likely to justify the grant of an exemption under Article 101(3) TFEU.

126.

In support of the opposing view, the applicants in the main proceedings argue that the restrictions at issue in the main proceedings are similar to certain restrictions which a licensor typically undertakes to give to a licensee. Such restrictions benefit from a block exemption if the market shares of the parties do not exceed certain thresholds or generally call for individual exemption even where those thresholds are exceeded.

127.

In particular, Roche emphasises that the restrictions whereby a licensor undertakes not to exploit licensed technology or not to sell, actively and/or passively, products incorporating that technology in the exclusive territory of the licensee or to an exclusive group of purchasers reserved to the licensee benefit from the block exemption provided for by Regulation No 772/2004 and by Regulation No 316/2014 which succeeded it. That is equally true, Roche alleges, whether the restrictions appear in an agreement between non-competing undertakings ( 79 ) or are inserted into an agreement between competing undertakings. ( 80 )

128.

It alleges that, even in cases where a block exemption cannot be granted because the applicable market-share thresholds are exceeded, according to the Guidelines, the said restrictions usually fulfil the conditions for individual exemption under Article 101(3) TFEU. ( 81 )

129.

I am unconvinced by that line of argument. Indeed, for the reasons which I set out in points 111 to 113 above, the restrictions at issue in the main proceedings cannot be reduced to the types of clauses, mentioned in the preceding points of this Opinion, to which those regulations and the Guidelines refer. That conclusion applies a fortiori if it is the case that the medicinal products concerned incorporate different technologies. Indeed, there could then be no question of the exploitation of licensed technology or the sale of products incorporating that technology.

130.

More generally, I doubt that restrictions such as those at issue in the main proceedings fall within the scope ratione materiae of those regulations, even in situations where, contrary to the AGCM’s findings in this case, the market-share thresholds beyond which block exemption may not be granted have not been exceeded. ( 82 )

131.

According to recital 9 of Regulation No 772/2004, so that the benefits of technology transfers may be attained, the regulation must cover provisions contained in technology transfer agreements that do not constitute the primary object of the agreement where they are ‘directly related to the application of the licensed technology’. Recital 9 of Regulation No 316/2014 states, more explicitly, that that regulation covers the provisions of such agreements only to the extent that they are ‘directly related to the production or sale of the contract products’. The restrictions at issue in the main proceedings, however, concern neither the production nor the sale of anti-VEGF medicines. They relate to the use and purchase of one of those medicines by parties unconnected with the licensing agreement relating to Lucentis.

132.

In light of all the foregoing considerations, I consider that the collusive conduct at issue does not fall outside the scope of Article 101(1) TFEU and does not benefit from exemption under Article 101(3) TFEU on the ground that the restrictions at issue in the main proceedings are similar to restrictions of competition that is exerted by a licensor with regard to a licensee that are included in a licensing agreement between non-competing undertakings.

133.

However, that conclusion does not decide the question of whether the collusive conduct at issue actually falls foul of the prohibition laid down in Article 101(1) TFEU. Nor does it imply that that conduct cannot be exempted under Article 101(3) TFEU once compliance with the conditions laid down in that provision has been individually analysed, ( 83 ) which is a matter for the parties who invoke that exemption to prove. ( 84 ) I shall deal with those aspects below, in the context of my analysis of the fifth question referred for a preliminary ruling.

134.

By its fifth question, the referring court asks whether collusion intended to ‘emphasise that [one] medicinal product is less safe or less efficacious’ than another may constitute a restriction of competition by object if, at the material time, there is neither ‘reliable’ scientific evidence to support the idea that that medicinal product is less efficacious or less safe, nor scientific knowledge to exclude that idea ‘indisputably’.

135.

In order better to focus my analysis, it seems to me that three preliminary points of clarification are needed in connection with the scope of that question in the light of the facts described in the order for reference.

136.

First of all, as is suggested by its wording, the fifth question is based on the premiss that, as the applicants in the main proceedings maintain, the equivalence of the safety and efficacy profiles of Avastin used off label and of Lucentis was the subject of scientific uncertainty at the time of the collusive conduct at issue.

137.

That premiss is disputed by the AGCM, Aiudapds, SOI-AMOI, the Region of Emilia-Romagna, Altroconsumo and the Italian Government. Those parties essentially argue that, while medical science is never such that the therapeutic equivalence of two medicinal products can be proven indisputably, the evidence available at the material time, which was subsequently corroborated by other evidence, ( 85 ) indicated the therapeutic equivalence of Avastin and Lucentis much more strongly than it called it into question. SOI-AMOI also emphasises that the safety and efficacy of Avastin in ophthalmology had already been demonstrated at the material time by means of a long-standing, worldwide medical practice. ( 86 )

138.

Given that it is not for the Court of Justice to call into question the factual context described by the national court, ( 87 ) my analysis of the fifth question must, I think, be based on that premiss. I shall therefore start from the premiss that the scientific debate concerning the therapeutic equivalence of the two medicines in question was, in any event, still ongoing.

139.

Secondly, the words ‘emphasise that a medicinal product is less safe or less efficacious’ may lead to confusion. I would prefer the neutral formulation of the ‘communication’ or ‘dissemination’ of ‘allegations’ in that regard.

140.

Indeed, the Italian word ‘enfatizzare’ that is used in the order for reference may also be translated into French by the expressions ‘mettre l’accent sur’ or ‘insister sur’, which do not suggest the exaggeration of information in terms of its content. ( 88 ) Moreover, as SOI-AMOI has observed, the exaggeration of or insistence on the lesser security or efficacy of one product by comparison with another presupposes that that lesser security or efficacy actually exists. The wording of the fifth question however indicates, on the contrary, that that lesser security or efficacy is the subject of a scientific debate. ( 89 )

141.

Thirdly, the order for reference and the case file submitted to the Court of Justice do not indicate that the AGCM criticised the applicants in the main proceedings for having disseminated not only opinions on the risks associated with the off-label use of Avastin but also allegations of the lesser efficacy of such use by comparison with the use of Lucentis.

142.

More precisely, the AGCM criticised the applicants in the main proceedings for having agreed upon a communication strategy to be adopted by Roche and Roche Italia with regard to the pharmaceutical regulatory authorities, medical practitioners and the general public. That strategy had allegedly consisted in insisting on the risks associated with the off-label use of Avastin and in spreading allegations of the lesser safety of that product by comparison with Lucentis. In particular, it was agreed that those companies would, on the basis of such allegations, request the EMA to amend the SPC for Avastin and to authorise the sending of a DHPC to ophthalmologists.

143.

In order to give a useful answer to the referring court, I shall therefore focus my analysis on an assessment of whether or not collusion in the communication of allegations of the lesser safety of one medicinal product compared to another is a restriction by object. ( 90 ) That said, the approach which I shall recommend on completion of that analysis will also apply to the concerted dissemination of allegations concerning both the comparative safety and the comparative efficacy of the two medicines.

144.

Bearing this in mind, I shall now address the question of whether, and if so to what extent, collusion in the communication to third parties of allegations of the lesser safety of one medicinal product used off label for certain therapeutic indications by comparison with a medicinal product that has been authorised for those indications constitutes a restriction of competition by object where the comparative safety of the two medicinal products is the subject of scientific uncertainty.

145.

According to settled case-law, the concept of ‘restriction of competition by object’ designates agreements and concerted practices which, in themselves, reveal a ‘sufficient degree of harm’ to competition to render the examination of their effects on competition superfluous. ( 91 )

146.

That case-law is based on the fact that ‘certain forms of coordination between undertakings can be regarded, by their very nature, as being harmful to the proper functioning of normal competition’. ( 92 )

147.

In order to establish whether particular collusive conduct is in the nature of a restriction by object, ‘regard must be had … to the content of its provisions, the objectives it seeks to attain and the economic and legal context of which it forms a part’. ( 93 ) That context also includes ‘the nature of the goods or services affected, as well as the real conditions of the functioning and structure of the market or markets in question’. ( 94 )

148.

In particular, this individual, detailed examination makes it possible to ‘understand the economic function and the real significance’ of the coordination at issue. ( 95 ) It also makes it possible to check whether there is a plausible alternative explanation for the coordination other than the pursuit an anticompetitive aim. ( 96 )

149.

In addition, the subjective intention of the parties, while not a necessary factor ( 97 ) or a sufficient factor ( 98 ) in order to find a restriction by object, may be a relevant factor for that purpose. ( 99 )

150.

I would add that, although the concept of restriction by object must be interpreted narrowly, ( 100 ) it is not limited to the forms of collusion expressly listed in Article 101(1) TFEU. ( 101 ) The atypical or novel form of a particular instance of collusion does not prevent the Court from concluding, after an individual, detailed examination, that that collusion, in itself, reveals a sufficient degree of harm to competition. ( 102 )

151.

Having regard to the principles I have outlined, and for reasons which I shall set out below, there can, in my view, be little doubt that collusive conduct concerning the dissemination of allegations of the lesser safety of one medicinal product by comparison with another does reveal a sufficient degree of harm to competition where those allegations are misleading (section (a) below). The aim of such conduct is to distort competition by exploiting a scientific uncertainty for the purpose of excluding the first medicinal product from the market or, at least, redirecting demand toward the second medicinal product.

152.

This first hypothesis corresponds to the version of the facts of the case presented to the Court by the AGCM, Aiudapds, SOI-AMOI, the Region of Emilia-Romagna, Altroconsumo and the Italian Government. Those parties argue, in substance, that the collusive conduct at issue involved the communication of allegations that did not reflect the state of scientific knowledge at the material time. ( 103 ) The aim of the conduct was to discourage the off-label use of Avastin in such a way as to alter demand in favour of Lucentis.

153.

On the other hand, where the allegations communicated are not misleading, such collusive conduct will not fall foul of the prohibition set out in Article 101(1) TFEU (section (b) below). In such a situation, the aim of the conduct is, in reality, to ensure the transparency of information regarding the safety of the medicinal products in question, so as to enable the recipients of such communications to take decisions of a kind that will protect public health. Such an aim promotes both public health and the free play of competition.

154.

This second hypothesis corresponds to the version of the material facts put forward by the applicants in the main proceedings. Spurred by genuine concerns about the safety of Avastin in ophthalmology, they merely exchanged information about the conduct that Roche and Roche Italia would adopt in order to fulfil their pharmacovigilance obligations. The applicants in the main proceedings add that the conduct had the more general aim of protecting public health and, in parallel with that, the reputation of the Roche group as manufacturer and distributor of Avastin. According to them, the concern was that the negative repercussions flowing from the risks associated with the off-label use of Avastin should not affect the on-label use of that medicinal product or the Roche group. ( 104 )

155.

Given that the examination of whether or not the allegations communicated were misleading calls for an appraisal of facts which falls within the exclusive jurisdiction of the referring court, it is for that court to choose between the various interpretations of the facts proposed by the interested parties and consequently to determine whether the collusive conduct at issue fits one or other of the two hypotheses I have described.

156.

To my mind, the concerted communication of misleading allegations of the lesser safety of one medicinal product compared to another is, by its very nature, harmful to the proper functioning of normal competition, so much so that an examination of its effects on competition is not necessary. ( 105 )

157.

First of all, where an examination of the content of the allegations in question reveals that they are misleading, the concerted communication of those allegations impairs the quality of the information available on the market and, consequently, adversely affects the decision-making process of those who create the demand for the two products concerned. Such concerted communication is, in itself, likely to reduce, if not suppress, demand for the first product to the advantage of the second.

158.

In my view, the communication of misleading allegations includes the dissemination of information which is in itself correct but is presented selectively or incompletely where, because of that manner of presentation, the information disseminated is likely to mislead those who receive it. ( 106 )

159.

To similar effect, Article 49(5) of Regulation No 726/2004 also provides that the holder of an AM may not communicate information relating to pharmacovigilance concerns to the general public without notifying the EMA and must, in any event, ensure that such information ‘is presented objectively and is not misleading’. ( 107 )

160.

That is the case whether or not there is continuing scientific uncertainty regarding the safety of a medicinal products. In my opinion, omitting to state that the risks created by using the medicine are uncertain, or exaggerating such risks with a lack of objectivity with regard to the available evidence, may render the concerted communication of those risks misleading. ( 108 )

161.

In the present case, the order for reference does not suggest that the content of the information regarding the adverse reactions caused by Avastin in ophthalmology, which the applicants in the main proceedings concerted to disseminate, was in itself incorrect. ( 109 ) The AGCM criticises them, in substance, for having presented that information incompletely and selectively, downplaying the value of scientific evidence to the contrary. Consequently, the allegations of the lesser safety of Avastin by comparison with Lucentis were lacking in objectivity and were therefore misleading.

162.

It will be for the referring court to determine, in the light of the considerations set out in points 158 to 160 above, whether or not those allegations were misleading, having regard to all the information that was at the disposal of the applicants in the main proceedings at the material time.

163.

Next, the objective of the concerted dissemination of misleading allegations of the lesser safety of one medicinal product by comparison with another is necessarily the exclusion of the first medicine to the advantage of the second, or at the very least a reduction in the demand for the first medicine. Given the misleading nature of such allegations, there can be no plausible alternative explanation for such collusion, in particular, one relating to the pursuit of legitimate aims concerning the transparency of the information available in the market and the protection of public health.

164.

If it is the case that the collusion at issue also pursued certain additional objectives unrelated to the restriction of competition, those may be taken into account only in the context of the possible application of Article 101(3) TFEU. ( 110 )

165.

In particular, the question could arise of whether the aim of halting the allegedly unlawful prescribing and marketing of Avastin for off-label use justifies the grant of an exemption under that provision.

166.

In this connection, I would mention in passing the guidance that may be drawn from the judgment in Slovenská sporiteľňa, ( 111 ) to which I have already referred, in which the Court analysed, with regard to Article 101(3) TFEU, an agreement to eliminate a competitor whose activities were supposedly unlawful (and which proved to be so after the agreement had been concluded). The Court left open the question of whether the removal of a competitor that is acting illegally could bring about gains in efficiency. In any event, the restriction of competition was not essential for achieving such gains in efficiency. The undertakings party to the agreement should have lodged a complaint against the competitor with the competent authorities, rather than take the law into their own hands and agree to exclude the competitor from the market. ( 112 )

167.

In my view, that reasoning also implies that, if the unlawfulness of prescribing or marketing a medicinal product for off-label use has not yet been made clear in a final decision of a competent court, ( 113 ) it is not for undertakings to assume that it is unlawful and act in concert to eliminate, by means of the dissemination of misleading opinions, the competitive pressure which those activities exert on the sale of another product.

168.

Finally, an examination of the economic and legal context, and in particular of the nature of the products and the operating conditions in the market in question, tends to confirm that collusion in the communication of misleading information regarding the lesser safety of one medicinal product by comparison with another is a restriction by object.

169.

As the AGCM, the Region of Emilia-Romagna, the French Government and the Commission have pointed out, medical practitioners are particularly sensitive to safety considerations surrounding medicinal products. Where those considerations concern the off-label use of a medicine, their aversion to risk is likely to increase in accordance with the legal provisions governing medical liability in force in the Member State in question. In this instance, according to the AGCM and the Italian Government, such liability does arise in Italy and is severe under civil and criminal law. Given that specific context, the widespread communication of alarming and misleading information about the risks associated with the off-label use of a medicinal product is inherently likely to discredit that product among doctors and stimulate demand for competing medicinal products.

170.

Moreover, the misleading nature of the opinions communicated, if established, will in itself mean that the collusive conduct at issue cannot be explained by the pursuit of the legitimate aims of ensuring the transparency of the information available in the market and the protection of public health and the reputation of the Roche group. That conclusion is all the more evident in that the attainment of such objectives did not necessitate concertation between the applicants in the main proceedings, having regard to the economic and legal context in which that conduct occurred.

171.

Indeed, while the manufacturer or MA holder (such as Roche) of a medicinal product (such as Avastin) must bear the risks, at least to its reputation, which arise from uncertainty regarding the off-label use of that medicine, such risks are in no way borne by other undertakings (such as Novartis) which market a competing medicinal product (such as Lucentis). It is not for the latter undertakings to help define the appropriate measures to attenuate the safety concerns associated with the off-label use of a medicinal product which they neither manufacture nor market. Similarly, as the AGCM, the Region of Emilia-Romagna, Altroconsumo and the Commission have argued, pharmacovigilance obligations fall solely upon the undertaking which holds the MA for the medicinal product in question.

172.

The subjective intention of the applicants in the main proceedings, as suggested by the findings of the AGCM that are set out in the order for reference, if established, could corroborate the existence of an anticompetitive aim behind the collusive conduct at issue. According to the AGCM, the applicants in the main proceedings expressed in various documents the intention to ‘generate and disseminate’ unjustified concerns relating to the safety of Avastin with a view to steering demand towards Lucentis. Thus, they sought to exploit uncertainty regarding the comparative safety of the two products in a manner favourable to their own commercial interests, but harmful to competition.

173.

I would add that, in the event that the referring court concludes that the allegations in question were misleading, a finding that the collusive conduct at issue was a restriction by object should be reached independently of the actual effects of that conduct.

174.

As other Advocates General have already emphasised, ( 114 ) and as the Court essentially clarified in its judgment in CB v Commission, ( 115 ) the individual, detailed examination of collusion should not be confused with the examination of its actual or potential effects on competition. If it were, the concepts of anticompetitive ‘object’ and of anticompetitive ‘effects’ would be merged, thus blurring the distinction which Article 101(1) TFEU draws between those two concepts. It is in that context that, according to the case-law, coordination may constitute a restriction by object if it has the ‘potential to’ have, or is ‘capable in an individual case’ of having an injurious effect on competition, without it being necessary to examine its actual effects. ( 116 )

175.

It is therefore immaterial, first of all, that the EMA refused to authorise the sending of a DHPC and made a different amendment to the SPC for Avastin from that requested by Roche. ( 117 ) Indeed, the fact that collusion does not meet with success in a given case is irrelevant to the identification of a restriction by object. ( 118 ) That fact may, however, be taken into account in the context of determining the amount of any fine. ( 119 )

176.

Secondly, the specialist competence of the pharmaceutical and ophthalmological regulatory authorities, which, according to the applicants in the main proceedings, enabled them to take a critical approach to the opinions communicated, is equally incapable of precluding a finding of restriction by object. On the contrary, I consider that, even if well-informed recipients have the necessary qualifications to foil a concerted strategy to disseminate misleading allegations about the safety of a product in order to reduce demand for it, the capacity of such a strategy to restrict competition cannot be called into question.

177.

The case of collusion relating to the communication of misleading allegations of the lesser safety of one medicinal product by comparison with another must be clearly distinguished from the case of concertation whereby the undertakings which hold the MAs for two medicines agree to communicate information about the comparative safety of those two medicines which, in light of scientific knowledge at the material time, is precise and objective.

178.

In my opinion, such concertation does not restrict competition within the meaning of Article 101(1) TFEU.

179.

Its aim, or its economic function and real meaning, is to improve the quality of the information available in the market so as to enable medical practitioners and pharmaceutical regulatory authorities to take enlightened decisions. Such an aim, as Roche emphasised at the hearing, promotes both the protection of public health and healthy competition. At the same time, the concerted communication of precise and objective information about the safety profile of a medicinal product enables the reputation of that medicine and of the undertakings which developed or manufactured it to be preserved.

180.

Concertation whereby the undertakings which hold the MAs for two medicinal products agree to communicate precise and objective information about the lesser safety of one of those medicines by comparison with the other is not, it seems to me, likely to produce anticompetitive effects.

181.

That conclusion follows logically from the counterfactual analysis which is required in order to identify a restriction of competition. Indeed, it is necessary to examine whether competition is restricted ‘within the actual context in which it would occur in the absence of the agreement in dispute’. ( 120 ) Concertation of this kind, rather that restricting the competition that would have existed in its absence, strengthens competition by ensuring the transparency of the information available in the market while at the same time endeavouring to protect public health.

182.

Consequently, if it is the case that the allegations which the applicants in the main proceedings concerted to disseminate were not misleading, the collusive conduct at issue would fall outside the scope of Article 101(1) TFEU.

183.

That is true despite the fact that the legitimate aims which I have just mentioned, relating to the transparency of information and the protection of public health and the reputations of Avastin and of the Roche group, could have been achieved unilaterally by the companies within that group. ( 121 )

184.

Admittedly, that fact affects the plausibility of the hypothesis of concertation aimed at achieving such legitimate objectives. It does not, however, render anticompetitive concertation to disseminate precise and objective information about the safety of a medicinal product. Once again, that conclusion follows from an examination of the situation which would have existed in the absence of the concertation. Indeed, assuming that the allegations at issue were not misleading, the actions taken by Roche and Roche Italia pursuant to the collusive conduct at issue would have been necessary even absent the collusion, so as to achieve the legitimate objectives I have mentioned and, in particular, to protect public health. ( 122 )

185.

I would add in this connection that, as the applicants in the main proceedings argue, the communication of precise, objective information about the safety profile of a medicinal product furthers the objectives which Regulation No 726/2004 pursues in instituting pharmacovigilance obligations. Notifying the pharmaceutical regulatory authorities of the presumed adverse reactions resulting from the off-label use of a medicinal product corresponds to the requirements laid down in Article 16(2) of that regulation and Article 104(1) of Directive 2001/83, to which Article 21(1) of the regulation refers. An application to amend the SPC of the medicinal product in question and for authorisation to issue a formal communication to medical practitioners, like the development of a strategy for communication to the general public, might in themselves constitute ‘appropriate measures’ to minimise any possible safety risks, in accordance with Article 104(2) of Directive 2001/83.

186.

It matters little that Regulation No 726/2004, like Directive 2001/83 did not extend the pharmacovigilance obligations to off-label uses of medicinal products until July 2012, ( 123 ) that is to say, after the collusive conduct at issue began. Undertakings cannot be criticised for adopting conduct consistent with those obligations, given that such conduct is in accordance with the intention of the legislature, guided by public health considerations.

187.

In light of all the foregoing considerations, I propose that the Court answer the questions referred by the Consiglio di Stato (Council of State, Italy) as follows: