Agreed by the Antimicrobial Advice ad hoc Expert Group (AMEG) 2 May Adopted by the CVMP for release for consultation 19 May 2016

Transcription

1 27 July 2016 EMA/CVMP/CHMP/231573/2016 Committee for Medicinl Products for Veterinry use (CVMP) Committee for Medicinl Products for Humn Use (CHMP) Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth Agreed by the Antimicrobil Advice d hoc Expert Group (AMEG) 2 My 2016 Adopted by the CVMP for relese for consulttion 19 My 2016 Adopted by the CHMP for relese for consulttion 23 My 2016 Strt of public consulttion 26 My 2016 End of consulttion (dedline for comments) 26 June 2016 Agreed by the Antimicrobil Advice d hoc Expert Group (AMEG) 1 July 2016 Adopted by the CVMP 12 July 2016 Adopted by the CHMP 22 July Churchill Plce Cnry Whrf London E14 5EU United Kingdom Telephone +44 (0) Fcsimile +44 (0) Send question vi our website An gency of the Europen Union Europen Medicines Agency, Reproduction is uthorised provided the source is cknowledged.

2 Tble of contents 1. Executive summry Introduction The use of colistin in humn nd veterinry medicine Humn medicine Veterinry medicine Antibcteril effect Resistnce mechnisms nd susceptibility testing Resistnce mechnisms Susceptibility testing Methodologicl pproches Monitoring results Possible links between the use of polymyxins nd other ntimicrobils in nimls nd resistnce in bcteri of niml origin Impct of use of colistin in food-producing nimls for niml nd humn helth Conclusions on updted literture review Profiling of the risk to public helth resulting from the use of colistin in nimls in the EU Hzrd identifiction Exposure Relese of resistnce genes from nimls treted with colistin Exposure of humns to resistnce genes vi bcteri from nimls Consequences to humn helth / hzrd chrcteristion Overll risk estimtion/chrcteristion Risk Mngement options Recommended risk mngement options for colistin in veterinry medicine Considertions when proposing risk mngement mesures Recommendtion on trget for use of colistin nd considertions on impct on use of other ntimicrobils Further considertions Justifiction for the trget Summry of the risk mitigtion recommendtions Strtegies for responsible use nd lterntives to the use of colistin Previously pplied risk mngement options New indictions, formultions or species Surveillnce of colistin consumption nd of colistin resistnce Generl considertions Follow up of the dvice Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 2/56

3 ANNEX Risk Mngement options tht were nlysed nd disregrded Withdrwl of existing mrketing uthoristions Group tretments Restriction on use for metphylxis Restriction from use in certin species Injectble, intrmmmry nd topicl formultions Figures Acknowledgement References Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 3/56

4 List of tbles Tble 1. Trends in consumption of polymyxins in EU/EEA countries, (expressed in DDD per inhbitnts nd per dy) Tble 2. Percentge of MDR isoltes in E. coli from poultry popultions nd met thereof, reported s resistnt to colistin Tble 3. Microbiologicl co-resistnce to colistin nd CIP nd/or CTX in E. coli from poultry popultions nd met thereof resistnce ssessed ginst ECOFFs (CST: MIC >2 mg/l, CIP: MIC >0.064 mg/l, CTX: MIC >0.25 mg/l) Tble 4. Clinicl co-resistnce to colistin nd CIP nd/or CTX in E. coli from poultry popultions nd met thereof resistnce ssessed ginst CBPs (CST: MIC >2 mg/l, CIP: MIC >1 mg/l, CTX: MIC >2 mg/l) Tble 5. Percentge of MDR isoltes in Slmonell spp. from poultry popultions nd met thereof, reported s resistnt to colistin Tble 6. Microbiologicl co-resistnce to colistin nd CIP nd/or CTX in Slmonell spp. from poultry popultions nd met thereof - resistnce ssessed ginst ECOFFs (CST: MIC >2 mg/l, CIP: MIC >0.064 mg/l, CTX: MIC >0.5 mg/l) Tble 7. Clinicl co-resistnce to colistin nd CIP nd/or CTX in Slmonell spp. from poultry popultions nd met thereof - resistnce ssessed ginst CBPs (CST: MIC >2 mg/l, CIP: MIC >1 mg/l, CTX: MIC >2 mg/l) Tble 8. Clssifiction of ntimicrobil clsses ccording to their probbility of trnsfer of resistnce genes nd resistnt bcteri Tble 9. Prevlence nd chrcteristics of mcr-1-positive isoltes from food-producing nimls, the environment, food nd humns, 1980s 2016 (updted from Skov & Monnet, 2016) List of figures Figure 1. Evolution of colistin use (J01XB01) in Belgin cute cre hospitls, (expressed in DDD per 1000 ptient-dys), strtified by type of cre (Primry = generl hospitls; Secondry = generl hospitl with teching missions; Tertiry = teching/university hospitl), modified from (Ingenbleek et l., 2015) Figure 2. Consumption estimtes bsed upon sles for food-producing nimls (including horses) of polymyxins, djusted for biomss under exposure (in mg/pcu), by country, for (EMA/ESVAC, 2015). No sles reported in Finlnd, Icelnd nd Norwy Figure 3. Distribution of veterinry sles for polymyxins by phrmceuticl form, djusted for biomss under exposure (in mg/pcu), by country for No sles in Finlnd, Icelnd nd Norwy. In ddition, negligible mounts were sold s bolus, orl pste, intrmmmries nd/or intruterine preprtions in some countries (EMA/ESVAC, 2015) Figure 4. Sles of colistin in for use in nimls in mg/pcu in 2013 (ESVAC dt), including the 5 nd 1 mg/pcu levels. No sles reported in Finlnd, Icelnd nd Norwy Figure 5. Sptil distribution of sles of polymyxins in veterinry medicine, in mg/kg biomss, in 26 EU/EEA countries, for No sles reported in Finlnd, Icelnd nd Norwy. (EMA/ESVAC, 2015). 38 Figure 6. Sptil distribution of sles of polymyxins in humn medicine, in mg/kg biomss, in 25 EU/EEA countries, for 2013 (dt shown only for countries reporting on totl consumption in the country; i.e. reporting for ntibiotic consumption in the community (outside hospitls) nd in the hospitl sector) (ECDC, 2015) Figure 7. Percentge of veterinry sles in mg/pcu for food-producing nimls, by phrmceuticl form of polymyxins, in the EU/EEA for No sles reported in Finlnd, Icelnd nd Norwy (EMA/ESVAC, 2015) (unpublished ESVAC dt 2013) Figure 8. Copy of the Februry 2016 cll for scientific dt for the updte of dvice Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 4/56

5 1. Executive summry Colistin is n ntibcteril gent of the polymyxin clss. Following the discovery of new colistin horizontlly trnsferble resistnce mechnism (MCR-1), the Europen Commission (EC) requested the Europen Medicines Agency (EMA) to updte the previous dvice on the impct of nd need for colistin use for humn nd niml helth (EMA, 2013). This updted dvice provides n nlysis of the colistin toxicity, susceptibility testing, ctivity nd resistnce mechnisms, risk profile (bsed upon the consumption ptterns nd epidemiology), nd risk mngement options. Soon fter its introduction in the 1950s, the use of colistin in humn medicine ws predominntly restricted to topicl dministrtions due to its toxicity if given systemiclly. Severe nosocomil infections due to multidrug-resistnt (MDR) Grm-negtive bcteri incresingly ccount for high morbidity nd mortlity nd colistin is therefore nowdys lst resort drug in humn medicine in the context of systemic tretment of infections cused by MDR Pseudomons eruginos, Acinetobcter bumnnii nd Enterobctericee (Escherichi coli, Klebsiell pneumonie). The prospect of novel lterntive ntimicrobils for tretment of infections due to MDR pthogens in the ner future is limited. The min indictions for systemic use in humn medicine re tretment nd control of infections in cystic fibrosis ptients nd tretment of severe systemic infections. In some countries orl colistin is in ddition used in prophylxis of helthcre-ssocited infections through selective digestive trct decontmintion (SDD). Totl consumption of colistin in humns (reflecting topicl, inhltionl nd systemic routes of dministrtion combined) vries widely between Europen Union/Europen Economic Are (EU/EEA) countries but hs doubled in some of EU/EEA countries between 2010 nd 2014 following the rise in MDR Grm negtive pthogens involved in helthcre-ssocited infections. Under routine lbortory conditions broth dilution methodology is recommended to determine colistin resistnce. Cre should be tken for proper identifiction to void overestimtion of cquired colistin resistnce due to some intrinsiclly less susceptible bcteri (Slmonell spp.) Bcteri contining ntimicrobil resistnce genes cn be selected through the use of colistin. Spred my be vi pssing on chromosoml genes to dughter colonies (verticl trnsmission) or vi mobile genetic elements (horizontl trnsmission). In isoltes from humns, colistin resistnce due to chromosoml mechnisms hs incresed drmticlly in some countries including Greece nd Itly but resistnce levels re now lso incresing in most other EU/EEA countries. Mobile (trnsferble) colistin resistnce, medited by the mcr-1 gene, hs been documented in severl EU/EEA countries. This is of gret concern due to the rpidly incresing use of colistin in EU/EEA hospitls leding to incresed selection pressure. Furthermore, other ntimicrobil clsses cn further stimulte the spred of colistin resistnce vi co-selection when there is simultneous presence of such resistnce genes (i.e. bet-lctmses, including crbpenemses). The mcr-1 gene ws found in similr plsmids in the sme bcteril species isolted from food-producing nimls, food, humns nd the environment indicting possible trnsmission between these comprtments. Nevertheless, the overll prevlence of colistin resistnce in nimls remins so fr nd with some exceptions low in food nd in nimls in the EU/EEA. Even though retrospective studies on collections of isoltes hve shown tht the mcr-1 gene hs been present in some bcteril species for decdes, recent dt from Chin (>20%) nd Jpn (13%) indicte tht the sitution is chnging rpidly nd tht the prevlence of such strins is incresing. The mcr-1 gene is present both in isoltes from clinicl cses of veterinry colibcillosis nd in invsive humn pthogens. Humn crriers cn become negtive within one month in the bsence of selection pressure. The reltive proportion mid humn clinicl isoltes in the EU/EEA remins firly low (less thn 1%), so fr. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 5/56

6 Colistin hs been used regulrly in veterinry medicine for decdes, both s curtive tretment nd for prevention of disese. It is of therpeutic importnce for the tretment of Grm-negtive gstrointestinl infections in certin food-producing species. Colistin is predominntly dministered s group tretment using the orl route of dministrtion. In 2013, polymyxins (minly colistin) were the 5th most sold group of ntimicrobils (6.1%) bsed on the totl sles of polymyxins in 26 EU/EEA countries reporting dt. The possible lterntives to colistin, depending on the resistnce sitution in prticulr country, include minopenicillins, trimethoprim, sulphonmides, tetrcyclines, minoglycosides, cephlosporins nd fluoroquinolones. If colistin is no longer vilble in veterinry medicine it could be speculted tht other ntimicrobils or mediction would replce its use if no concomitnt interventions such s vccintion or improved biosecurity mesures re tken. The more frequent isoltion of the mcr-1 gene in veterinry isoltes compred to humn isoltes up until the present time (Tble 9), together with the much higher use of colistin in livestock compred to humn medicine nd the finding of the mcr-1 gene, long with genetic determinnts typiclly seen in niml environments, hs been considered suggestive of flow from nimls to humns. In December 2014 the CVMP recommended to restrict the indictions for use of colistin to tretment of enteric infections cused by susceptible non-invsive E. coli only, tht ny indictions for prophylctic use should be removed nd tht the tretment durtion should be limited to the minimum time necessry for the tretment of the disese nd not exceed 7 dys. In ddition, it ws recommended to remove horses from the Summry of Product Chrcteristics (SPCs) on the grounds of trget species sfety concerns. In April 2016 the CVMP recommended the withdrwl of the mrketing uthoristions for ll veterinry medicinl products for orl use contining colistin in combintion with other ntimicrobil substnces. There is wide vrition between Europen Union (EU) Member Sttes (MS) in the extent of veterinry use of colistin. From the dt vilble the vrition cnnot be directly linked to the predominnce of specific niml species, ctegory or husbndry system in n individul MS, with some MS hving low level or no use of the substnce, suggesting tht there is scope to decrese the overll use of colistin within the EU. Antimicrobil use in both humn nd veterinry medicine must be rtionlised nd reserved for clinicl conditions. Further to previous dvice, the Antimicrobil Advice d hoc Expert Group (AMEG) min recommendtions, which were endorsed by the CVMP nd the CHMP re tht colistin sles for use in nimls should be reduced to the minimum fesible (see below) nd tht colistin should be dded to higher risk ctegory (ctegory 2) of the AMEG clssifiction (EMA, 2014). There re wide vritions in the use of colistin djusted for the biomss under exposure (kg livestock, expressed s popultion correction unit (PCU)) 1, between countries nd these re lrgely unexplined. Countries with intensive livestock production cn hve level of usge below 1 mg/pcu (e.g. Denmrk nd the UK) or much higher, up to 20 to 25 mg/pcu (Itly nd Spin). Considering the rpidly incresing importnce of colistin for tretment of criticlly ill humn ptients, ll countries should strive to reduce the use of polymyxins s much s possible. For the current "high nd moderte consumers" the trget nd desirble levels re set t 5 mg/pcu nd 1 or below 1 mg/pcu, respectively, bsed on the observtions on the level of use in other countries. Menwhile more informtion should be gthered to determine the minimum level of colistin 1 The popultion correction unit (PCU) corresponds to the food-producing niml popultion tht cn be subject to tretment with ntimicrobil gents, for further detils see: 0b01c &jsenbled=true Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 6/56

7 use tht cn be chieved while mintining niml welfre nd preventing the incresed use of other Criticlly Importnt Antimicrobils (CIAs). Reduction in use of colistin should be chieved without n increse in the use (in mg/pcu) of fluoroquinolones, 3rd- nd 4th-genertion cephlosporins or overll consumption of ntimicrobils. The bove trgets for reduction in sles of colistin should be chieved in period of three to four yers. If the sitution regrding colistin resistnce in nimls or humns further deteriortes, it my be necessry to lower the proposed trgets. 2. Introduction The globl emergence nd stedy increse in bcteri tht re resistnt to multiple ntimicrobils hs become public helth thret (Crlet et l., 2012). Humn infections with MDR bcteri re ssocited with higher ptient morbidity nd mortlity, higher costs nd longer length of hospitl sty (Cosgrove, 2006; Huck et l., 2016; Schorr, 2009). In the current stte of incresing resistnce coupled with decrese in the vilbility of new ntibiotics, there is need to explore ll options tht would llow, s fr s possible, the preservtion of the current ntimicrobil rmmentrium (ECDC/EMEA, 2009). Colistin (polymyxin E) is ctionic, multicomponent lipopeptide ntibcteril gent tht ws isolted by Koym et l from the broth of Penibcillus (Bcillus) polymyx vr. colistinus in the lte 1940s (Koym et l., 1950). The ntimicrobil ws used cliniclly in nimls in the 1950s nd in humns in the 1960s. In humn medicine, colistin ws erly on predomintely restricted to topicl use due to its systemic toxicity (Nord nd Hoeprich, 1964). The lst 10 yers, incresing numbers of hospitl outbreks with crbpenemse-producing Enterobctericee (E. coli, Klebsiell species), nd MDR Pseudomons nd Acinetobcter species (i.e. non-fermenttive Grm-negtive bcteri), hve forced clinicins to re-introduce systemic colistin tretment, s lst resort drug for the tretment of helthcre-ssocited infections in which these orgnisms re involved. Colistin therefore incresingly hs key role for public helth, despite ll the limittions deriving from its sfety profile nd uncertinties round the best wy of using it (Ntion nd Li, 2009). Also, colistimethte sodium (CMS) is used by inhltion for the tretment of P. eruginos lung infections in ptients with cystic fibrosis. In certin countries prophylxis of helthcre-ssocited infections by mens of SDD lso includes the use of colistin in the ntimicrobil regimen. Colistin hs been used for decdes in veterinry medicine, especilly in swine nd vel clves. Bsed on SPCs (prior to the lst referrl procedures, see chpter 3.2. for further detils) Grm-negtive infections of the intestinl trct, due to E. coli nd Slmonell spp. were the primry indictions. Most of the colistin pplictions in nimls re for orl group tretments. In July 2013 the AMEG ws convened on behlf of the Europen Commission by the Europen Medicines Agency nd concluded tht for colistin use in prticulr, detiled monitoring of colistin resistnt bcteri is required to confirm horizontl gene trnsfer is not involved nd tht overll prevlence remins low. As soon s colistin resistnce determinnts re found on mobile genetic elements in the bcteri of concern s well s from humn or niml origin, or clonl explosion of virulent bcteri tkes plce, new risk ssessment would be required (EMA, 2013). Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 7/56

8 In light of this recommendtion, nd following the recent discovery of mcr-1, horizontl trnsferble resistnce gene in bcteri of food niml origin (Liu et l., 2015), the impct of the current or future use of colistin products in veterinry medicine for niml helth nd welfre hs been re-ssessed. 3. The use of colistin in humn nd veterinry medicine 3.1. Humn medicine Due to the mjor concerns for neuro- nd nephrotoxicity (Koch-Weser et l., 1970; Ryn et l., 1969), prenterl use of polymyxins hs until recently been limited nd polymyxins were minly used for ophthlmic nd topicl infections (Flgs nd Ksikou, 2005; Koch-Weser et l., 1970). Cystic fibrosis ptients hve been n exception to this prctice for decdes, nd such ptients hve received systemic or nebulised colistin to control lower irwy bcteril infections nd their complictions (Beringer, 2001; Tppenden et l., 2013). During the lst five yers two mjor indictions hve renewed the interest for polymyxins in humn medicine, nmely s prt of surgicl prophylxis vi SDD nd for the tretment of MDR Grm-negtive helthcre-ssocited infections. For humn ptients, two slt forms of polymyxin E (colistin) hve been widely commercilly vilble, nmely colistin sulphte nd colistimethte sodium (CMS, syn colistin methnesulphte, colistin sulphonyl methte, pentsodium colistimethnesulphte). CMS is prodrug of colistin nd is microbiologiclly inctive (Bergen et l., 2006). It is dministered predominntly s prenterl formultions nd vi nebulistion (Flgs nd Ksikou, 2005). After dministrtion, CMS is hydrolysed to colistin, which is the bse component tht is responsible for its ntibcteril ctivity (Lim et l., 2010). Colistin sulphte is vilble in tblets nd syrup for SDD nd s topicl preprtions for skin infections. CMS is vilble for dministrtion intrvenously, intrmusculrly s well s topiclly vi erosol (nebulistion) or intrventriculr dministrtion. Besides polymyxin E (colistin), polymyxin B is lso widely used in humn medicine. Although prenterl formultions exist nd re used in vrious prts of the world, in the EU/EEA polymyxin B is used only for topicl dministrtion in humns. Outside of the EU polymyxin B is vilble in prenterl formultions nd cn be dministered intrvenously, intrmusculrly, or intrthecl. Helthcre-ssocited infections cused by multi drug-resistnt (MDR) Grm-negtive orgnisms re being incresingly reported, especilly in ptients hospitlised in intensive cre units nd hemtology/oncology units (Zrb et l., 2012). Colistin hs re-emerged s lst resort therpeutic option to tret infections due to MDR, lctose-fermenting nd -non-fermenting Grm-negtive bcilli, including P. eruginos nd A. bumnnii, for which there is growing unmet medicl need. In prticulr, clinicins nowdys incresingly hve to resort to colistin to tret nosocomil infections in criticlly ill ptients, such s bcteremi nd ventiltor-ssocited pneumoni (VAP), due to crbpenem-resistnt Grm-negtive bcteri (Dikos et l., 2012; Petrosillo et l., 2013). In most cses these crbpenem-resistnt orgnisms produce serine-bsed crbpenemse (e.g. the KPC or OXA enzymes) (Cnton et l., 2012) or metlloenzyme (e.g. the New Delhi Metllo-β-Lctmse 1, NDM-1 nd the Veron integron-encoded metllo-β-lctmse, VIM) (Bogerts et l., 2010; Corngli et l., 2011; Kumrsmy et l., 2010). These bcteril strins pper to be spreding within the EU nd hve become mjor problem in some centres/countries (ECDC, 2016; Hung et l., 2011). Colistin is co-dministered with other ntibiotics such s tigecycline or crbpenems in some countries s limited vilble tretment options for crbpenemse-producing Enterobctericee, Acinetobcter spp. nd Pseudomons spp. (Dikos et l., 2012; Qureshi et l., 2012; Tumbrello et l., 2012). A recent rndomised tril filed to estblish clinicl benefit for the combintion of colistin with Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 8/56

9 rifmpicin for the tretment of serious infections due to extremely drug-resistnt (XDR) A. bumnnii, despite synergism ws shown in vitro (Durnte-Mngoni et l., 2013). The use of colistin by inhltion s djunctive therpy or s monotherpy for tretment of VAP hs lso been explored (Lu et l., 2012; Michlopoulos nd Flgs, 2008; Rttnumpwn et l., 2010); lrger rndomised trils re needed in order to conclude on the utility of this pproch. Avilble evidence, minly from old cse series, suggests tht systemic colistin is n effective nd cceptbly sfe option for the tretment of children without cystic fibrosis who hve MDR Grm-negtive infections (Flgs et l., 2009). For MDR nd XDR Grm-negtive infections, recent survey mong 94 children hs found colistin to be non-inferior to non-colistin tretment group (Ozsurekci et l., 2016), lthough in both groups infection-relted mortlity ws high (11% nd 13.3%, respectively). The mjor dverse effects of the systemic use of colistin in humns re nephrotoxicity (cute tubulr necrosis), nd neurotoxicity such s presthesi, dizziness/vertigo, wekness, visul disturbnces, confusion, txi, nd neuromusculr blockde, which cn led to respirtory filure or pnoe (Flgs nd Ksikou, 2005). Older studies show much higher frequency of neurotoxicity nd occsionlly irreversible nephrotoxicity (pproximtely 7%), compred to more recent studies. The exception is cystic fibrosis ptients in whom up to 29% dverse (neurologicl) effects hve been reported (Bosso et l., 1991; Reed et l., 2001). The need for higher doses of CMS to chieve dequte colistin concentrtions for therpeutic effect, s shown in recent studies (Gronzik et l., 2011; Plchours et l., 2009), rises concerns round the consequent further increse in nephrotoxicity (Pogue et l., 2011). To contin toxic side-effects following systemic use of colistin, close monitoring of renl function nd voidnce of co-dministrtion with other nephrotoxic gents (e.g. minoglycosides) re recommended (Flgs nd Ksikou, 2005). New derivtives of polymyxins, with more fvourble toxicity profile re under evlution (Vr nd Vr, 2013). The use of prenterl colistin to tret serious humn infections ws hmpered in the pst by remining uncertinties regrding the optimum dose regimen, by the use of different wys to describe nd express the dose (in milligrms colistin bse nd s Interntionl Units) nd by the uncertinty regrding wht is ctully delivered s ctive substnce to the ptient (Gronzik et l., 2011; Mohmed et l., 2012; Vicri et l., 2013). In the context of recent rticle 31 (of Directive 2001/83/EC) referrl procedure, the EMA Committee for Humn Medicinl products (CHMP) reviewed the existing evidence nd decided to revise the pproved indictions so tht colistin cn be used without ge restrictions, but only for the tretment of infections with limited tretment options. The posology nd method of dministrtion section of the Summry of Product Chrcteristics (SmPC) were revised, nd the need of loding dose ws greed upon. No firm recommendtions could be nevertheless mde for ptients with heptic or renl impirment nd for ptients on renl replcement therpy, due to the scrcity of dt for these subpopultions (EMA, 2014b) 2,3. Within the sme frmework, the CHMP lso reviewed the optiml wy of expressing the strength nd dose of colistin nd greed tht the EU product informtion for CMS will continue to be expressed in Interntionl Units (IU). At the sme time, dose content conversion tble between CMS (expressed in IU nd in mg) nd colistin bse ctivity (expressed in mg) ws introduced to help the prescribers. Colistin is used in humn medicine both in the community nd hospitl sectors, nd there is growing need in specific settings like intensive cre units (Ingenbleek et l., 2015) nd for tretment of Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 9/56

10 helthcre-ssocited infections due to crbpenemse-producing Grm-negtive bcteri (ECDC, 2016). Medicl doctors often hve to rely on colistin for the tretment of these infections. Alterntive ntibcterils such s tigecycline, fosfomycin nd temocillin lso hve limittions nd re sometimes uthorized only in limited number of countries cross EU MSs. Few new ntimicrobils for systemic infections with MDR Grm-negtive pthogens re expected in the future. Of notice, new bet (β-)lctm- β-lctmse inhibitor combintion product (ceftzidime-vibctm), which is ctive ginst orgnisms tht produce serine-bsed but not metllo-bsed crbpenemses, ws pproved by the Food nd Drug Administrtion of the USA (FDA) in 2015 nd received positive opinion from the CHMP in April Totl consumption (reflecting topicl, inhltionl nd systemic routes of dministrtion combined) vries widely between EU/EEA countries nd doubled between 2010 nd 2014 (ECDC, 2015) following the rise in MDR Grm-negtive pthogens involved in helthcre-ssocited infections (Skov nd Monnet, 2016). Tble 1 shows the distribution of nd trends in the consumption of polymyxins (minly colistin) for systemic use in EU/EEA countries. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 10/56

11 Tble 1. Trends in consumption of polymyxins in EU/EEA countries, (expressed in DDD per inhbitnts nd per dy) Source: Europen Centre for Disese Prevention nd Control (ECDC): Summry of the ltest dt on ntibiotic consumption in the Europen Union, ESAC-Net surveillnce dt, November 2015 (ECDC, 2015) Country Trends in consumption of polymyxins, Averge nnul chnge Sttisticl significnce Finlnd (b) n.. Lithuni () n.. Norwy <0.001 significnt Polnd () n.. Ltvi <0.001 n.s. Sweden <0.001 n.s. Netherlnds n.s. Bulgri <0.001 n.s. Estoni <0.001 < <0.001 n.s. Denmrk <0.001 n.s. Luxembourg <0.001 n.s. Sloveni n.s. United Kingdom ()(d) n.. Hungry significnt Frnce <0.001 n.s. Mlt n.s. EU/EEA <0.001 n.s. Irelnd <0.001 n.s. Portugl (c) n.s. Croti n.s. Slovki () n.. Itly significnt Greece () n.. Belgium n.. The number for EU/EEA refers to the corresponding popultion-weighted men consumption, clculted by summing the products of ech country s consumption in DDD per inhbitnts n per dy x country popultion s in Eurostt, nd then dividing this sum by the totl EU/EEA popultion. ) These countries did not report dt for ll yers during the period b) Finlnd: dt include consumption in remote primry helthcre centres nd nursing homes. c) Portugl: dt relte to public hospitls only. d) United Kingdom: dt do not include consumption from UK-Wles (2013) or UK-Northern Irelnd (2014). n..: not pplicble; liner regression ws not pplied due to missing dt. n.s.: not significnt. Long-term, detiled surveillnce is needed to monitor the evolution t the country level nd strtified by specility. For exmple in Belgium, the use of colistin hs more thn doubled in intensive cre units ccording to the ltest surveillnce dt, in prticulr in university hospitls (Figure 1). Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 11/56

12 Figure 1. Evolution of colistin use (J01XB01) in Belgin cute cre hospitls, (expressed in DDD per 1000 ptient-dys), strtified by type of cre (Primry = generl hospitls; Secondry = generl hospitl with teching missions; Tertiry = teching/university hospitl), modified from (Ingenbleek et l., 2015) Evolution is expressed in DDD (defined dily dose) per 1000 ptient-dys for hospitl wide non-peditrics wrds (left) nd intensive cre units (right). Prticiption rtes exceed on verge >85% mong 110 cute cre hospitls over consecutive yers. Virulent clones of K. pneumonie or other difficult to tret Grm-negtive bcteri re becoming resistnt during therpy nd ssocited with hospitl outbreks within the EU/EEA nd worldwide (Blm et l., 2013; Brink et l., 2013; Comndtore et l., 2013; Del Bono, 2013; Lmbrini, 2013; Lesho et l., 2013; Monco et l., 2014; Onori et l., 2015; Snitkin et l., 2013). Anlysis of nosocomil outbreks with A. bumnnii indicted tht prior crbpenem nd colistin consumption my hve cted s triggering fctors for the development of resistnce (Agodi et l., 2014; Wright et l., 2016). As outlined below, the mcr-1 gene hs now been shown in different humn isoltes including invsive pthogens both in hospitl nd mbultory cre (Tble 9) (Meletis et l., 2011), nd outbreks due to MDR pthogens expressing the mcr-1 gene might occur in the ner future. Colistin resistnce thus hs been emerging rpidly following its reintroduction for prenterl use in humn medicine, s shown in different reports, with n ssocited incresed mortlity (Cpone et l., 2013; Kontopoulou et l., 2010; Zrkotou et l., 2010). In hospitl in Greece, colistin resistnce rtes rose from 0% in 2007 to 8.1% in 2008 nd to 24.3% in 2009 (Meletis et l., 2011). The ltest estimtes from Itly show rise of colistin resistnce in K. pneumonie from 1 to 2% in 2006 to 33% in 2009 (Monco et l., 2014). Prior to the discovery of the mcr-1 gene, Dutch survey hs demonstrted tht colistin resistnce, shown to be clonl in nture fter orl use in the ICU for SDD, cn rpidly spred in hospitl nd therefore SDD should be discourged in outbrek settings (Hlby et l., 2013). Since mcr-1-producing bcteri lredy hve been isolted from limited number of humn ptients (Tble 9) Poirel et l. (2016) expressed similr concerns nd requested n urgent review of SDD, given the occurrence of horizontlly trnsferble colistin resistnce. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 12/56

13 3.2. Veterinry medicine Within the EU MSs, colistin nd polymyxin B re uthorised ntionlly. The min indiction for colistin in veterinry medicine is infection of the gstrointestinl trct cused by non-invsive E. coli in pigs, poultry, cttle, sheep, gots nd rbbits. Colistin is lso used in lying hens nd cttle, sheep nd gots producing milk for humn consumption. Colistin is lso ctive ginst endotoxins produced by some E. coli strins in the gstrointestinl trct. Typiclly, colistin products re dministered orlly, in feed, in drinking wter, s drench, or through milk replcer diets. Combintions of colistin with other ntimicrobils re vilble for group tretments of food-producing nimls in some EU countries. Products for prenterl nd intrmmmry dministrtion re lso vilble, nd infections due to Grm-negtive bcteri in ruminnts including endotoxemi re climed indictions. Polymyxin B is on the list of substnces essentil for the tretment of equide for systemic tretment for endotoxemi (ntitoxigenic effect, not ntibcteril s such) ssocited with severe colic nd other gstrointestinl diseses (Brton et l., 2004; Moore nd Brton, 2003; Officil Journl of the Europen Union, 2013). As in humn medicine, colistin nd polymyxin B hve been registered for topicl dministrtion to individul veterinry ptients, except for food-producing nimls in the cse of polymyxin B, in the bsence of MRLs. In compnion nimls, prescription eye nd erdrops re vilble with colistin lone, or in combintion with other ntimicrobils. Colistin tblets re vilble for clves for the prevention nd tretment of neontl colibcillosis. In some EU MSs, veterinry medicinl products (VMPs) contining colistin re not on the mrket, i.e. not commercilised (EMA/ESVAC, 2015). Colistin products (polymyxin E) hve never been mrketed for use in nimls in the United Sttes (US Food nd Drug Administrtion, 2016). Sources from the FDA hve indicted tht there is only one polymyxin B product (ophthlmic ointment, combintion of polymyxin B nd oxytetrcycline) pproved for use in food-producing species. In recent yers, this product hs been mrketed in 2009 nd , lthough it hs been mrketed in smll quntities. Polymyxin B is lso vilble in the US s component of pproved ophthlmic products (for use in dogs nd cts) nd otic products (for use in dogs). There is documented legl off-lbel use in other non-food-producing species, such s horses. Sources from the Public Helth Agency of Cnd hve indicted tht there re no pproved colistin products (polymyxin E) for use in nimls in Cnd (Public Helth Agency of Cnd, 2016). In the EU/EEA, colistin hs been used in veterinry medicine since the 1950s (Koym et l., 1950), primrily for pigs including group tretments nd prevention of dirrhoe cused by E. coli nd Slmonell spp., s first choice tretments for neontl dirrhoe in piglets (Timmermn et l., 2006) nd vel clves (Prdon et l., 2012) cused by E. coli s well s for the therpy of mild colibcillosis in poultry. The medin number of individuls treted with colistin per 1000 nimls nd per dy in Belgium ws 41.3 (Cllens et l., 2012b) nd 58.9 (Prdon et l., 2012) for finishing pigs (50 frms) nd for vel clves (15 frms), respectively. Bsed on the overll ntimicrobil consumption, these studies demonstrte tht colistin ccounted for more thn 30% of the ntimicrobil use in swine nd 15% in vel frming. The Belgin use of colistin ws for indictions others thn those for which it is uthorised, e.g. respirtory disese, peritonitis (Prdon et l., 2012) nd streptococcl infections (Cllens et l., 2012b). Doses vried between niml species, frm types nd indictions. Timmermn (2006) reported underdosing (sub-dosing) of orl colistin in piglets possibly due to dilution in food or wter, since its dministrtion ws not weight-bsed. Studies on diry frms hve shown limited use of polymyxins (Ctry et l., 2016; Ctry et l., 2007; Menéndez González et l., 2010). In 32 broiler frms in Belgium, the use of colistin ws not reported despite detiled ntimicrobil consumption records (Persoons et l., 2012), lthough colistin hs been used in medicted feed ( Older studies from in limited number of Belgin cttle frms, Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 13/56

14 hve shown tht feed (strter rtions) with ntibiotics were given for 6 to 13 dys in ll of 5 exmined vel clves frms nd 55% of them contined colistin (Ctry et l., 2007). In the sme survey nd in gret contrst, the men number of suckling beef (n= 5 frms) nd diry cttle (n= 5 frms) tht received colistin ws on verge below 0.2 per 1000 nimls dily (Ctry et l., 2007). In 2013, the totl sles of polymyxins in the 26 EU/EEA countries reporting dt to the ESVAC project, including tblets but excluding topicl forms, polymyxins were the 5th most sold group of ntimicrobils (6.1%), fter tetrcyclines (36.7%), penicillins (24.5%), sulphonmides (9.6%), nd mcrolides (7.4%) (Figure 7). Totl sles in weight summed up 495 tonnes. Of those 99.7% were for orl forms s follows: 43.3% were orl solution (powder nd liquid for use in drinking wter), 42.5% were premix (premixes for medicted feeding stuff) nd 14.0% were orl powder (powder to be dministered with the feed or milk). Smll mounts were sold s: injectbles (0.2%), tblets (0.1%) nd intrmmmries, intruterines nd orl pste (less thn 0.0% for ech of the three forms). Of the group of polymyxins, colistin represented more thn 99.9% of the sles. In ddition combintions of colistin with other ntimicrobils re uthorised in some MSs. The sles of those combintion products represents less thn 10% of the overll sles of colistin (dt not published). Some MSs with high consumption of polymyxins hve shown decrese in consumption between 2011 nd 2013, wheres others hve shown stble sitution or even n increse (Figure 2). In Belgium, polymyxin use showed 28.1% decrese in This reduction seen for the second yer in row hs been ttributed due to strt of the use of zinc oxide s n lterntive for colistin use in the tretment of post-wening dirrhoe in piglets (BelVetSc, 2015). The lst ESVAC report shows n overll decrese of 19% of sles of polymyxins in 23 countries over the lst yer (EMA/ESVAC, 2015). Colistin is used in quculture for the prevention of Grm-negtive infections (Xu et l., 2012), consumption dt re not vilble seprtely for this food production sector. In the Dnish monitoring progrmme (DANMAP), detils on consumption do not refer to the use of colistin in fish (DANMAP, 2012). Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 14/56

15 Figure 2. Consumption estimtes bsed upon sles for food-producing nimls (including horses) of polymyxins, djusted for biomss under exposure (in mg/pcu), by country, for (EMA/ESVAC, 2015). No sles reported in Finlnd, Icelnd nd Norwy Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 15/56

16 Figure 3. Distribution of veterinry sles for polymyxins by phrmceuticl form, djusted for biomss under exposure (in mg/pcu), by country for No sles in Finlnd, Icelnd nd Norwy. In ddition, negligible mounts were sold s bolus, orl pste, intrmmmries nd/or intruterine preprtions in some countries (EMA/ESVAC, 2015). Due to concerns tht the differences in posology nd withdrwl periods estblished cross the EU for veterinry medicinl formultions contining colistin t IU per ml nd intended for dministrtion in drinking wter to food-producing species could present potentil serious risk to public nd niml helth, the United Kingdom referred the mtter to the Agency on April 2009, under Article 35 of Directive 2001/82/EC, s mended (EMA/CVMP, 2010). In their opinion the CVMP confirmed tht the benefit risk blnce remined positive for the use of colistin for tretment of gstrointestinl infections cused by non-invsive E. coli susceptible to colistin, when dministered t dose of IU colistin per kg body weight dily for clves, lmbs, pigs nd IU colistin per kg body weight dily in poultry for 3-5 consecutive dys. The risk-benefit blnce regrding the use of colistin for tretment of gstrointestinl infections cused by Slmonell spp. in clves, lmbs, pigs nd poultry ws considered negtive, nd those indictions were removed from the SPCs of the involved products. The scope of the mentioned referrl ws limited to veterinry medicinl products contining colistin for dministrtion in drinking wter; products dministered in feed (or injectbles) were not ddressed. Subsequent to the AMEG s previous dvice in 2013, further referrl ws concluded under Article 35 of Directive 2001/82/EC for ll VMPs contining colistin s sole substnce dministered orlly (including premixes) to food-producing nimls (EMA/CVMP, 2015). In December 2014 the CVMP recommended to restrict the indictions for use of colistin to tretment of enteric infections cused by susceptible non-invsive E. coli only, tht ny indictions for prophylctic use should be removed nd the tretment durtion should be limited to the minimum time necessry Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 16/56

17 for the tretment of the disese nd not exceeding 7 dys. In ddition, it ws recommended to remove horses from the SPCs on the grounds of trget species sfety concerns. In April 2016 the CVMP recommended the withdrwl of the mrketing uthoristions for ll veterinry medicinl products for orl use contining colistin in combintion with other ntimicrobil substnces Antibcteril effect The bctericidl effect of colistin is the result of n electrosttic interction with divlent ctions of the outer bcteril membrne, which cuses disruption of the cell structure, lekge of the cell contents nd thereby cell lysis (Lim et l., 2010; Schindler nd Osborn, 1979). The brod-spectrum of ctivity of polymyxins ginst Grm-negtive bcteri involves binding to lipid A, the nchor for lipopolyscchride, nd the min constituent of the outer membrne of these bcteri. Time kinetic-kill in vitro studies hve shown concentrtion-dependent bctericidl ction (Guyonnet et l., 2010). Polymyxins re produced nturlly by Bcillus (Penibcillus) polymyx. Polymyxins re prticulrly ctive ginst wide rnge of species of Grm-negtive bcilli (e.g. E. coli, Slmonell spp. nd P. eruginos) including those displying crbpenem resistnce, nd certin Mycobcterium species. Colistin differs from polymyxin B, only by one mino cid in position 6 (D-leucine in colistin, phenyllnine in polymyxin B). Both compounds hve the sme mechnism of ction nd resistnce development. Polymyxin B nd colistin (sulphte) hve similr spectrum of ntibcteril ctivity ginst min Grm-negtive pthogens (Gles et l., 2011). Polymyxins hve no cliniclly useful ctivity ginst Grm-positive bcteri, Grm-negtive cocci, nerobes nd Mollicutes including Mycoplsm spp. (Flgs nd Ksikou, 2005). In ddition, colistin lcks therpeutic ctivity ginst intrinsiclly (inherently) resistnt species, including bcteri of the gener Serrti, Stenotrophomons, nd Proteus spp. (Pogue et l., 2011). Colistin heteroresistnce, (i.e. cultures where both susceptible nd resistnt subpopultions re present), hs been reported for K. pneumonie (Poudyl et l., 2008), P. eruginos (Bergen et l., 2011), A. bumnnii nd E. cloce (Hwley et l., 2008; Lo-Ten-Foe et l., 2007). The potentil for under-dosing in reltion to selecting subpopultions with higher MICs, during tretment with colistin hs been illustrted for A. bumnnii (Dvid nd Gill, 2008). The use of combintion therpy would hve the potentil benefit to reduce the emergence of such subpopultions. Studies tht included moth (Glleri mellonell) infection model hve found tht vncomycin nd doripenem might hve synergistic effect together with colistin in A. bumnni strins with decresed colistin susceptibility (O'Hr et l., 2013). For P. eruginos, synergistic effects hve been shown in vitro between colistin nd mny other compounds (e.g. rifmpicin nd the nti-pseudomonl gents zlocillin, pipercillin, ztreonm, ceftzidime, imipenem, doripenem, or ciprofloxcin) (Conwy et l., 1997). Recent studies hve demonstrted tht colistin is synergistic with drugs of the echinocndin fmily ginst Cndid species, by incresing permebilistion nd ttck by colistin on fungl membrnes (Zeidler et l., 2013). The phrmcokinetic/phrmcodynmic (PK/PD) pproch hs been pplied successfully to the selection of dose regimens for new ntibcteril gents nd the re-evlution of efficcious dose regimens for severl ntimicrobil clsses. PK/PD hs some potentil to identify regimens tht my minimise selection pressure for resistnt strins. Although the vst mjority hve focused on the prevention of muttionl resistnce (Drlic nd Zho, 2007), some studies hve shown benefit for the continment of bcteri in which resistnce is medited minly by horizontl gene trnsfer (McKinnon et l., 2008). The ppliction of PK/PD for colistin hs only recently re-gined ttention due Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 17/56

18 to its incresing systemic use to tret multidrug-resistnt bcteri cusing humn infections. The PK/PD prmeter to mximise bctericidl ctivity nd minimise resistnce hs been shown s the re under the inhibitory curve (AUIC, or fauc/mic) for trget orgnisms such s P. eruginos nd Acinetobcter spp. (Michlopoulos nd Flgs, 2011). In veterinry medicine, similr estimtes hve been found to be relible for preclinicl studies for colibcillosis in piglets (Guyonnet et l., 2010). 4. Resistnce mechnisms nd susceptibility testing 4.1. Resistnce mechnisms Acquired resistnce to colistin in normlly susceptible bcteri hs for long been chrcterised by chromosoml muttions nd thus in theory ws non-trnsferble by mobile genetic elements (Cllens et l., 2012b; Lndmn et l., 2008; Olitn et l., 2014). Chromosoml polymyxin resistnce is medited by muttions in specific regions (pmra/b nd phop/q)(moskowitz et l., 2012). Resistnce is then ssocited with chnges in the trget components of the Grm-negtive bcteril wll, nmely covlent ddition of 4-mino-L-rbinose (LAr4N) to phosphte groups within the lipid A nd oligoscchride s elements from the lipopolyscchride (LPS) (Boll et l., 1994; Moskowitz et l., 2012; Moskowitz et l., 2004; Nummil et l., 1995). The two-component regultory PrR-PrS system with n identicl modifiction of LPS is involved in the dptive resistnce t sub-inhibitory concentrtions of ctionic peptides, including colistin nd the bovine peptide, indolicidin (Fernndez et l., 2010). Reserch hs demonstrted tht the ctivity of lysozyme nd other innte immune defence peptides (LL37) cn be ffected (Npier et l., 2013). Colistin resistnce thus confers resistnce to polymyxins nd rnge of other ctionic peptides. Decresed ctivity of polymyxins is due to structurl LPS chnges t both the cytosol nd peri-plsmtic site of the cell membrne (Moskowitz et l., 2012). Studies indicte similr (temperture dependent) mechnism in other bcteri including A. bumnnii, Yersini enterocolitic nd Slmonell spp. (Beceiro et l., 2011; Beceiro et l., 2011b; Guo et l., 1997; Reines et l., 2012). They found tht the development of moderte level of colistin resistnce in A. bumnnii requires distinct genetic events, including (i) t lest one point muttion in pmrb, (ii) up-regultion of pmrab, nd (iii) expression of pmrc, which leds to the ddition of phosphoethnolmine to lipid A (Beceiro et l., 2011). The phop/q system hs been shown to be involved in strins with intrinsic resistnce, for exmple pthogenic Edwrdsiell trd from fish (Lv et l., 2012) nd K. pneumonie (Wright et l., 2015). These systems re different from the mechnisms of colistin resistnce in lbortory nd clinicl strins of A. bumnnii s described by (Mofftt et l., 2010), whom noted unexpectedly the totl loss of LPS production vi inctivtion of the biosynthesis pthwy genes lpxa, lpxc, or lpxd. In Yersini spp., polymyxin resistnce cn be relted to the existence of efflux pumps with potssium nti-porter systems (RosA/RosB) (Bengoeche nd Skurnik, 2000). In K. pneumonie muttions in crrab, present in mny multidrug resistnt virulent strins (ST258, see below), histidine kinse gene s prt of two-component regultory system (TCRS), hve been found involved in decresed colistin susceptibility (Wright et l., 2015). Colistin-resistnt mutnts of E. coli, K. pneumonie, A. bumnnii nd P. eruginos cn be selected in vitro from cultures progressively grown in medium contining 0.5 to 16 µg/ml colistin (Lee et l., 2016). With the exception of some well-exmined clinicl strins (K. pneumonie), mny of the bove muttion mechnisms re not stble fter severl pssges in vitro (Moskowitz et l., 2012). This instbility of polymyxin resistnce by muttion, hs been for long nd prior to the discovery of the mcr-1 gene, stted to reduce the risk of rpid spred of resistnce to colistin (Gentry, 1991; Lndmn Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 18/56

19 et l., 2008). Investigtions on consecutive smples of A. bumnnii from nosocomil infections hve indicted tht this in vitro instbility of colistin resistnce is lso found in vivo during colistin therpy (Lesho et l., 2013; Snitkin et l., 2013; Yoon, 2013). Out of 37 ptients treted with colistin for less thn one to three months, in five ptients (13%) muttions in the pmr locus were found. Colistin susceptibilities returned soon fter cesstion of colistin therpy (Snitkin et l., 2013), but in one of the isoltes n pprently more stble muttion ws found (pmrb L271R ). Of note is tht this strin s grdient diffusion (E-test) nd microbroth dilution susceptibility tests were highly discordnt (Snitkin et l., 2013). Proteomic nlysis by Chu nd collegues hve shown tht low intrcellulr c-di-gmp concentrtions in bcteri (i.e. secondry messenger required for dpttions in life style of bcteri) re ssocited with polymyxin resistnce. Biofilm formtion by bcteri, which hs long been regrded s leding to decresed susceptibility to ntimicrobils, is systemticlly down-regulted t low intrcellulr c-di-gmp concentrtions (Chu et l., 2013). Biofilms re protective lyers round bcteri tht re formed, for exmple, round inert invsive devices (e.g. implnts) or in the digestive trct s mucosl biofilm communities (Fite et l., 2013). Wheres for mny ntimicrobil gents, resistnce trnsfer is enhnced under biofilm conditions, this down-regultion of c-di-gmp might explin why this is not pplicble for colistin resistnce. In other words, colistin resistnce, nd mybe by extension colistin presence, might interfere with biofilm formtion nd therefore resistnce trnsfer. To wht extent conjugl deficiency nd down-regultion of biofilm formtion re relted within the occurrence of colistin resistnce, is not documented. An exhustive updte on chromosoml colistin resistnce mechnisms (verticl trnsmissible) ws done by Olitn et l. (2014). In the 1980 s, work on K. pneumonie did indicte tht colistin-resistnt mutnts counterct horizontl gene trnsfer from multi-resistnce gene clusters (Lmousin-White nd O'Cllghn, 1986). This conjugl deficiency of colistin-resistnt strins ws found to be 1000-fold compred to colistin-susceptible strins under lbortory conditions. No lter reports hve confirmed these findings nd underlying mechnisms. This spect of colistin-resistnt isoltes hs been nevertheless t tht time exploited successfully under clinicl circumstnces. Although stepwise muttionl resistnce hs ppered following prolonged colistin use in certin hospitl outbreks, becuse plsmids were not present in the epidemic strins, the colistin-resistnt isoltes remined susceptible to other ntibiotics. Through the rottionl use of colistin nd minoglycosides, the prevlence of resistnt Klebsiell spp. decresed during the ltter outbreks (O'Cllghn et l., 1978). More recently genomic nlysis hve suggested possible fitness cost due to colistin resistnt muttions with loss of β-lctmse-encoding plsmids (Wright et l., 2016). Wheres some mcr-1 hrbouring plsmids do not show so fr identified resistnce genes (Suzuki et l., 2016), mny E. coli hrbour β-lctmses together with mcr-1 (Tble 9) including decresed susceptibility for crbpenems s well s resistnce determinnts for other ntimicrobil clsses (Poirel et l., 2016). In November 2015, Liu et l. (2015) reported tht trnsferble plsmid-medited colistin resistnce gene, mcr-1, hd been found in E. coli isoltes from nimls, food nd bloodstrem infections from humn ptients in Chin. Subsequent retrospective nlysis of strin collections showed the mcr-1 gene ws lredy circulting in the 1980 s (Shen et l., 2016) nd the EU/EEA in vriety but low bsolute number of Grm-negtive orgnisms (Doumith et l., 2016). Although the exct mechnism is under exmintion, the mcr-1 gene encodes membrne-nchored phosphoethnolmine trnsferse tht likely confers resistnce to colistin by modifying lipid A (Thnh et l., 2016). The mcr-1 gene is often ssocited with trnsposble elements locted on different types of plsmids (phnshp45, IncI2, IncX4, IncHI2 nd IncP2 ) (Liu et l., 2015; Thnh et l., 2016; Zeng et l., 2016). These plsmids hve been shown to hve high in vitro trnsfer rtes (10 1 to 10-2 ) or bsent, Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 19/56

20 depending on the conditions nd strins involved. Conjugtion hs been shown from E. coli nd Slmonell spp. into other Enterobctericee, not only K. pneumonie, Enterobcter erogenes nd Enterobcter spp. but lso P. eruginos. (Cllens et l., 2016; Quesd et l., 2016; Zeng et l., 2016). Linked resistnce genes hve been shown in mny isoltes (Tble 9). The MIC observed in strins crrying mcr-1 hs rnged from 0.5 to 32 mg/l nd is stted to be ssocited with the diversity of lipid A structures found in Enterobctericee (Thnh et l., 2016). The mcr-1 positive E. coli strins cn hve other colistin resistnce genes due to muttions in chromosoml DNA present (PmrA/B), nd of notice these strins filed to trnsfer the mcr-1 gene in conjugtion mting experiments (Quesd et l., 2016). The occurrence of the mcr-1 gene in E. coli nd lso cross different slmonell serovrs hs been recently confirmed in different EU MSs like Belgium (Botteldoorn, submitted), Spin (Quesd et l., 2016), the Netherlnds (Veldmn, 2016), nd Frnce (Perrin-Guyomrd et l., 2016) with specil relevnce for turkeys Susceptibility testing Methodologicl pproches Susceptibility testing of colistin is performed by testing colistin sulphte since the prodrug CMS is completely inctive s shown by Bergen et l. (2006) nd ll its ctivity seen in vitro simply would derive from prtil conversion of CMS to colistin over time. In the lst couple of yers there hs been intensive reserch under the uspices of Europen Committee on Antimicrobil Susceptibility Testing (EUCAST) nd Clinicl nd Lbortory Stndrds Institute (CLSI) to delinete methods tht could produce relible nd reproducible susceptibility results. Presently only broth dilution cn be recommended for susceptibility testing, i.e. for the time being neither disk diffusion, gr dilution nor grdient test should be used for testing of colistin. Broth microdilution (BMD) should be performed using uncoted polystyrene microtiter pltes; ction djusted Mueller-Hinton broth without ny other dditives (in prticulr no polysorbte 80 or other surfctnts) (EUCAST homepge The EUCAST clinicl brekpoints for Enterobctericee (E. coli nd Klebsiell spp., but excluding Proteus spp., Morgnell morgnii, Providenci spp., nd Serrti spp.), P. eruginos, nd A. bumnnii re 2 µg/ml for colistin susceptible isolte; nd >2 µg/ml for colistin resistnt isolte (EUCAST, 2013). For non-clinicl surveillnce purposes, the epidemiologicl cut-off vlue (ECOFF) cn be difficult to determine given certin slmonell serovrs, such s Dublin nd Enteriditis demonstrte subpopultions tht re (intrinsiclly) slightly-less susceptible (Agersø et l., 2012). A number of new techniques for susceptibility testing nd identifiction of resistnce determinnts hve been developed (Jung et l., 2014; O'Neill, 2015; Vn Belkum nd Dunne, 2013). These techniques reduce the ntimicrobil susceptibility testing time from two to four dys to pproximtely one to two hours, which could reduce the empiricl tretment nd stimulte pproprite ntimicrobil use. The utility of colistin resistnce determintions hs recently been demonstrted for E. coli (Liu et l., 2016), with method clled SERS-AST (simple surfce-enhnced Rmn ntimicrobil susceptibility testing). For the interprettion of Tble 9, it is of importnce to stress tht in the bsence of reserch into the specificity nd sensitivity of the mcr-1 gene, PCR (test chrcteristics identifying flse positive/negtive results), nd estimtion of the true (bsolute prevlence) prevlence is difficult. In prticulr only isoltes with elevted MICs ccording to the ltest EUCAST/CLSI recommendtions might hve been included. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 20/56

21 Monitoring results Occurrence of microbiologicl resistnce to colistin A summry of n extrction of ll vilble phenotypic dt on colistin resistnce from the Europen Union summry report on ntimicrobil resistnce in zoonotic nd indictor bcteri from humns, nimls nd food in 2014 (EFSA/ECDC, 2016) is given here: Twenty fourteen ws the first yer of mndtory EU monitoring for colistin resistnce in Slmonell spp. nd indictor E. coli from nimls. Although some MSs encountered technicl difficulties in ccurtely determining colistin susceptibility, the monitoring dt obtined re being considered to bseline in poultry (niml species trgeted for 2014) ginst which future chnges cn be mesured. The reported occurrence of colistin resistnce is unlikely to equte directly to the occurrence of the mcr-1 gene, becuse number of different resistnce mechnisms cn confer colistin resistnce s indicted in previous section of this report. In the cse of Slmonell spp., dt were reported nd is presented for broilers, lyers, fttening turkeys, met from broilers nd met from turkeys. For E. coli dt were reported nd is only vilble for broilers nd fttening turkeys. The ECOFF vlue pplied for the nlysis of the occurrence of microbiologicl resistnce to colistin in both Slmonell spp. nd E. coli ws >2 mg/l. EU hrmonised monitoring dt indicted tht 0.9% of E. coli from broilers (totl tested equl to 4037, colistin-resistnce found in 24 MSs) nd 7.4% of E. coli from fttening turkeys (totl tested equl to 1663, colistin-resistnce found in 11 MSs) were colistin-resistnt ccording to the interprettive criteri pplied. In the cse of Slmonell spp., 8.3% of isoltes from broilers (totl tested=1683, colistin-resistnce found in 10 MSs), 2% of isoltes from fttening turkeys (totl tested=757, colistin-resistnce found in 6 MSs), 14.1% of isoltes from lying hens (totl tested=822, colistin-resistnce found in 13 MSs), 24.7% of isoltes from turkey met (totl tested equl to 279, colistin-resistnce found in 2 MSs), nd 4.4% of isoltes from broiler met (totl tested equl to 911, colistin-resistnce found in nine MSs) were colistin-resistnt ccording to the interprettive criteri pplied. Resistnce ws detected in diversity of slmonell serovrs, lthough lrge proportion of the colistin-resistnt Slmonell spp. from broilers nd lying hens were S. Enteritidis. There re studies showing tht the distribution of the wild type differs between serovrs. A generl ECOFF vlue therefore cn led to flse positive resistnce interprettion for some serovrs or subpopultions herein (Agersø et l., 2012) Multidrug resistnce in colistin resistnt isoltes Dt on multidrug resistnce in E. coli isoltes from poultry popultions nd met thereof, reported in the EU from hrmonised surveillnce s resistnt to colistin re presented in Tble 2. In this nlysis we included the E. coli isoltes originting from lying hens, broilers, nd fttening turkeys flocks; nd isoltes from broilers nd turkey met, for which ntimicrobil resistnce (AMR) dt to the following 12 ntimicrobils were reported: mpicillin (AMP), cefotxime (CTX), ceftzidime (CAZ), nlidixic cid (NAL), ciprofloxcin (CIP), tetrcycline (TET), gentmicin (GEN), trimethoprim (TMP), sulphonmide (SUL), chlormphenicol (CHL), meropenem (MERO) nd colistin (CST). For the purpose of this nlysis, resistnce to CIP/NAL nd CTX/CAZ hve been ddressed together. Dt on microbiologicl nd clinicl co-resistnce to colistin nd in ddition to criticlly importnt ntimicrobils (CIP nd/or CTX) in E. coli from poultry popultions nd met thereof re presented in Tble 3 nd Tble 4. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 21/56

22 Tble 2. Percentge of MDR isoltes in E. coli from poultry popultions nd met thereof, reported s resistnt to colistin N Res. colistin Res 0 Res 1 Res 2 Res 3 Res 4 Res 5 Res 6 Res 7 Res 8 Res % 2.6% 1.2% 1.2% 6.2% 11.1% 13.0% 25.9% 32.1% 8.6% 0.6% 0% N: totl number of E. coli isoltes from poultry origin nd met derived thereof tested ginst nine clsses of ntimicrobils; Res 0: number (%) of isoltes resistnt to colistin only nd to none of the nine dditionl ntimicrobil clsses. Res 1-Res 9: number (%) of isoltes resistnt to colistin being lso resistnt to one ntimicrobil clss/resistnce to nine ntimicrobil clsses. Tble 3. Microbiologicl co-resistnce to colistin nd CIP nd/or CTX in E. coli from poultry popultions nd met thereof resistnce ssessed ginst ECOFFs (CST: MIC >2 mg/l, CIP: MIC >0.064 mg/l, CTX: MIC >0.25 mg/l) N Res. colistin Not Res. to CIP nor CTX Res. to CIP or CTX Res. to both CIP nd CTX (2.6%) 33 (20.4%) 120 (74.1%) 9 (5.6%) N: totl number of E. coli isoltes from poultry origin nd met derived thereof tested ginst nine ntimicrobil clsses. Tble 4. Clinicl co-resistnce to colistin nd CIP nd/or CTX in E. coli from poultry popultions nd met thereof resistnce ssessed ginst CBPs (CST: MIC >2 mg/l, CIP: MIC >1 mg/l, CTX: MIC >2 mg/l) N Res. colistin Not Res. to CIP nor CTX Res. to CIP or CTX Res. to both CIP nd CTX (2.6%) 87 (53.7%) 73 (45.1%) 2 (1.2%) N: totl number of E. coli isoltes from poultry origin nd met derived thereof tested ginst nine ntimicrobil clsses. Dt on MDR, in slmonell isoltes from poultry popultions nd met thereof, reported in the EU s resistnt to colistin re presented in Tble 5. Dt on microbiologicl nd clinicl co-resistnce to colistin nd in ddition to criticlly importnt ntimicrobils (CIP nd/or CTX) in Slmonell spp. from poultry popultions nd met thereof re presented in Tble 6 nd Tble 7. In this nlysis we included the Slmonell spp. isoltes originting from lying hens, broilers, nd fttening turkeys flocks; nd isoltes from broilers nd turkey met, for which ntimicrobil resistnce dt to the following 12 ntimicrobils were reported: AMP, CTX, CAZ, NAL, CIP, TET, GEN, TMP, SUL, CHL, MERO nd colistin. For the purpose of this nlysis, resistnce to CIP/NAL nd CTX/CAZ hve been ddressed together. Tble 5. Percentge of MDR isoltes in Slmonell spp. from poultry popultions nd met thereof, reported s resistnt to colistin N Res. colistin Res 0 Res 1 Res 2 Res 3 Res 4 Res 5 Res 6 Res 7 Res 8 Res % 8.5% 62.6% 26.8% 1.3% 3.2% 3.5% 2.1% 0.5% 0% 0% 0% N: totl number of Slmonell spp. isoltes from poultry origin nd met derived thereof tested ginst nine clsses of ntimicrobils; Res 0: number (%) of isoltes resistnt to colistin only nd to none of the nine dditionl ntimicrobil clsses. Res 1-Res 9: number (%) of isoltes resistnt to colistin being lso resistnt to one ntimicrobil clss/resistnce to nine ntimicrobil clsses. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 22/56

23 Tble 6. Microbiologicl co-resistnce to colistin nd CIP nd/or CTX in Slmonell spp. from poultry popultions nd met thereof - resistnce ssessed ginst ECOFFs (CST: MIC >2 mg/l, CIP: MIC >0.064 mg/l, CTX: MIC >0.5 mg/l) N Res. colistin Not Res. to CIP nor CTX Res. to CIP or CTX Res. to both CIP nd CTX (8.5%) 309 (82.0%) 67 (17.8%) 1 (0.3%) N: totl number of Slmonell spp. isoltes from poultry origin nd met derived thereof tested ginst nine ntimicrobil clsses. Tble 7. Clinicl co-resistnce to colistin nd CIP nd/or CTX in Slmonell spp. from poultry popultions nd met thereof - resistnce ssessed ginst CBPs (CST: MIC >2 mg/l, CIP: MIC >1 mg/l, CTX: MIC >2 mg/l) N Res. colistin Not Res. to CIP nor CTX Res. to CIP or CTX Res. to both CIP nd CTX (8.5%) 373 (98.9%) 4 (1.1%) 0 (0%) N: totl number of Slmonell spp. isoltes from poultry origin nd met derived thereof tested ginst nine ntimicrobil clsses. 5. Possible links between the use of polymyxins nd other ntimicrobils in nimls nd resistnce in bcteri of niml origin Despite the bundnt use of colistin in veterinry medicine for over 50 yers, retrospective nlysis of bcteril collections showed tht trnsmission of colistin resistnce in Grm-negtive bcteri vi horizontl gene trnsfer or sustined clonl expnsion hs not been substntil in the EU/EEA. Following the first Asin reports, confirmtion of the mcr-1 gene in lrge dtbses in UK (Doumith et l., 2016) mong 15 out of isoltes of Slmonell spp., E. coli, Klebsiell spp. Enterobcter spp. nd Cmpylobcter spp. from food nd humn isoltes from between 2012 nd 2015 hs been done, while the number of reports is ever growing in the EU/EEA nd worldwide (Tble 9). In the ltest reports, mcr-1-postive isoltes from clinicl specimens so fr remin uncommon (Cnntelli et l., 2016). To dte the erliest niml isoltes were in the 1980s in Chin nd were from poultry (Shen et l., 2016); the erliest humn isolte ws Shigell sonnei strin in 2008 from Vietnm (Skov nd Monnet, 2016; Thnh et l., 2016). More reserch is needed becuse of the diversity of plsmids nd occurrences of the mcr-1 gene in different ecosystems including surfce wter (Tble 9). The lrger bundnce in veterinry isoltes compred to humn cses, together with the by fr exceeding quntities of colistin use in livestock (ECDC/EFSA/EMA, 2015) hs been considered suggestive of flow from nimls to humns (Skov nd Monnet, 2016). Nordmnn & Poirel (2016) recently hve listed further rguments for this rtionle side from the difference in colistin use nd resistnce prevlence. First, the occurrence of isoltes with simultneous resistnce for florfenicol which is only uthorised for used in nimls (Poirel et l., 2016), nd the co-presence of extended-spectrum β-lctmses typicl of niml origin, CMY-2 (Flgenhuer et l., 2016). Homologies in the genetic orgnistion of mcr-1 with insertion sequences in n importnt ubiquitous niml pthogen P. multocid (Poirel et l., 2016). Given tht the mcr-1 gene is present in isoltes tht often hrbour other resistnce determinnts like those encoding β-lctmse production (Tble 9), co-selection of these isoltes by other ntimicrobils thn polymyxins should be considered. A review of ntimicrobil consumption in livestock t lrge is therefore provided in the next prgrphs. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 23/56

24 Low ntimicrobil consumption is found in diry nd beef cttle tht hve regulr ccess to psture. Under these conditions, 5-10 nimls re treted on verge with stndrd ntimicrobil dose per 1000 nimls (equl to tretment incidence; TI), for colistin the TI ws found to be lower thn 0.2/1000 (Ctry et l., 2007). For grzing nimls, resistnce in E. coli is low for most ntimicrobils, but multi-resistnce is encroching slowly over consecutive yers (Geenen et l., 2011; MARAN, 2012). In vel clves in centrl Europe, the verge overll TI with ntimicrobils ws clculted to be 417 per 1000 nimls per dy (Prdon et l., 2012), nd for colistin this dily incidence is pproximtely 60 per The evolution of multi-drug resistnce is worrisome in vel clves (MARAN, 2012), yet colistin resistnce in this production system hs historiclly been extremely low to bsent (Di Lbio et l., 2007). Ltest findings hve however demonstrted the presence of mcr-1 in clinicl isoltes from vel (Henni et l., 2016; Mlhotr-Kumr et l., 2016). The ltest figures from Belgium show grdul decrese in colistin resistnce in E. coli from vel clves, from 14.7% in 2011 to 6.7% in 2014 (CODA-CERVA, 2015). In Belgium, the second highest ntimicrobil-consuming livestock production system is tht of fttening pigs, where on verge over 200 to 250 per 1000 individuls re treted dily with ntimicrobils (Cllens et l., 2012b). Up to 30% of orl prophylctic nd metphylctic group tretments consist of colistin (Cllens et l., 2012b). If pproprite testing is pplied, resistnce is only recent, but incresingly (10% in Belgium) being reported mong porcine pthogenic E. coli strins (Boyen et l., 2010). In commensl E. coli from Belgin pigs nd with the exception of very slight increse in 2013, colistin resistnce is considered very low over the period (CODA-CERVA, 2015). Dutch, porcine E. coli nd slmonell isoltes, s reported in 2009 (MARAN, 2009), remin fully susceptible. Lrge studies combining consumption nd resistnce re limited, becuse colistin susceptibility tests s routinely performed re not fully relible or vilble. A lrge surveillnce study in Polish livestock reveled 0.9% of E. coli (n=1728) to be resistnt to colistin for the period (Wsyl et l., 2013). In centrl Europen broilers, pproximtely 95 to 130 nimls were reported to be treted dily with stndrd ntimicrobil dose per 1000 individuls (MARAN, 2009; Persoons et l., 2012). Quntifiction of broiler consumption did not identify use of colistin in 50 rndomly selected frms in Belgium (Persoons et l., 2012), but it is used in mny other EU MSs. The Dutch MARAN report covering 2009 showed decrese in the use of intestinl nti-infectives (including colistin nd neomycin) in broilers from 26.0 to 18.4 dily dosges per 1000 nimls (conversion from dily dosges per niml yer). Colistin resistnce in E. coli from broilers is incresingly becoming ssocited with multi-resistnce (Geenen et l., 2011). Nevertheless, reports of colistin resistnce remin scrce nd limited to some broiler met smples (2.1%, N=328) (MARAN, 2009) nd more recently turkey (4.5%) (MARAN, 2015). A retrospective study from the Netherlnds demonstrted presence of 10% mcr-1 in E. coli from turkey met (Veldmn, 2016). In Itly, Bttisti nd coworkers found high prevlence while screening turkey isoltes (E. coli, Slmonell spp. from monitoring). In the non-selective monitoring, prevlence of mcr-1 in E. coli from fttening turkeys ws 22%, nd in isoltes from ESBL-screening 25% (Bttisti, 2016b). A recent report from Germny hs reveled tht, in prticulr, turkey nd turkey-derived food (6-18%) frequently contined colistin-resistnt E. coli compred to broilers nd broiler derived food (2-8%) (Alt et l., 2015). Cre should be tken tht technicl difficulties cn result in over-reporting of colistin resistnce, in prticulr for Slmonell spp. when contminted with inherent resistnt orgnisms such s Proteus species. Studies on ntimicrobil consumption nd further processed in the production chin of turkeys should be done in the future to investigte the Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 24/56

25 resons for the reltive high prevlence of colistin resistnce, prticulrly in turkey nd met thereof compred with other production types. In Austrlin Aeromons strins from fish hve frequently been found to hve decresed susceptibility to colistin (55.5%), especilly when retrieved from clinicl cses (Arven-Romn et l., 2012), lthough this might be intrinsiclly present. Studies under EU/EEA quculture conditions re not vilble. Surveillnce dt until 2014 show low levels of colistin resistnce despite considerble colistin use especilly in vel nd fttening pigs (Cllens et l., 2012) with even decrese or low stedy stte during the lst couple of yers in Belgium (Hnon et l., 2015), nd Sweden (Swedres-Svrm, 2014). Detiled ccurte monitoring is needed in these confined production systems to follow up the emergence of clonlly resistnt strins nd to demonstrte bsence of multi-resistnce plsmids or lterntive structures tht include efficient spreding mechnisms for polymyxin resistnce. In Chin (Shen et l., 2016), Tiwn (Kuo et l., 2016), nd Frnce (Perrin-Guyomrd et l., 2016) the occurrence of mcr-1 from food-producing nimls shows n increse of colistin resistnce during the most recent yers which might be of importnce for prediction of potentil for the further globl spred (Grmi et l., 2016). The Netherlnds (SD, 2015) nd Belgium (BelVetSc, 2015) hve set nd ttined trgets to reduce the consumption of ntimicrobils in veterinry medicine over limited number of yers. In the Netherlnds for instnce, 58% (50% in fttening pigs) hs been demonstrted over the period from 2009 to Along, decrese of overll resistnce in fecl bcteri hs been found in E. coli in livestock in the Netherlnds (MARAN, 2015). In Belgium, fter two consecutive yers of substntil reduction in consumption djusted for kg biomss in 2012 (-6.9%) nd 2013 (-6.3%), disppointing results were found for 2014 (+1.1%) (BelVetSc, 2015). A decrese in resistnce in indictor E. coli from different Belgin livestock species hs lso been found (CODA-CERVA, 2015). An increse in Chinese livestock production (broilers, i.e. chicken rised for met, nd swine) by nerly 5% in upcoming yers ( ) is nticipted s is subsequent increse in colistin use (Liu et l., 2015). Doses given for growth promotion outside the EU/EEA cn be severl times lower thn the doses given for metphylxis nd curtive purposes to EU/EEA livestock, nd subsequent concerns for different selection pressure of mcr-1 hve been rised (Richez nd Burch, 2016). A lrge retrospective nlysis showed the presence of this gene in the erly 1980s in Chin, nd rther quickly fter the use of colistin in niml production (Shen et l., 2016). In the EU/EEA detils on the chronology nd occurrence of mcr-1 in nimls nd wys of dministrtions re lcking to investigte to wht extent differences in selection pressure hve n impct on the occurrence nd spred of the mcr-1 gene. From the retrospective nlysis of dtbnks worldwide so fr (Tble 9), it is cler tht trnsferble colistin resistnce ws out there but only detected within weeks, nd highest prevlence hve been demonstrte only in the most recent yers of interests. Bsed upon the prevlence of colistin resistnce nd mcr-1 in turkey or turkey met in prticulr (Bttisti, 2016b; Perrin-Guyomrd et l., 2016; Veldmn, 2016), e.g. from 0 in 2007 to 6% in 2014 in French turkey isoltes, detiled investigtions in this livestock production sector on colistin consumption nd ntimicrobils t lrge re lcking to demonstrted ssocitions with these findings. 6. Impct of use of colistin in food-producing nimls for niml nd humn helth Colistin is now regrded s lst line defence ginst infections cused by MDR Grm-negtive bcteri such s K. pneumonie nd A. bumnnii. Its clinicl use hs resurged in mny prts of the Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 25/56

26 world despite the limittions posed by its toxicity profile. The use of colistin in combintion is more frequently considered nd clinicl studies re on-going. Humn nosocomil infections with colistin-resistnt strins, prticulrly with crbpenem resistnt K. pneumonie, with high mortlity hve been reported (Cpone et l., 2013; Kontopoulou et l., 2010; Zrkotou et l., 2010). The only independent risk fctor demonstrted for colistin-resistnt, crbpenemse-producing Enterobctericee (CPE) in mtched, controlled studies, is the use of colistin itself (Brink et l., 2013; Hlby et l., 2013). Often encountered in the EU/EEA is K. pneumonie sequence types (ST) 258, resistnt to ll bet (β)-lctms, cephlosporins, crbpenems (KPC/clss A; non-metllo), fluoroquinolones, mcrolides, minoglycosides, tigecycline, nd colistin (Comndtore et l., 2013; Dhr et l., 2016). This colistin-resistnt vrint of ST258 is circulting widely in Greece, with clinicl cses lso seen, possibly vi importtion, in Hungry, the UK (Livermore, 2012) nd USA (Bogdnovich et l., 2011). Other multi-resistnt exmples re K. pneumonie ST 14 nd ST17, reported in Asi (Blm et l., 2013). Despite the presence of mny other horizontlly-trnsferble extended spectrum resistnce mechnisms (e.g. β-lctms nd crbpenems), the colistin resistnce determinnts remin locted on the chromosome nd do not pper to be horizontlly trnsferble. It is cknowledged tht, s shown for the clone ST258 (Bogdnovich et l., 2011), these strins hve high cpbility for successful spred. In EU/EEA livestock, enteric diseses re treted with colistin, minly in swine nd poultry. The mount of colistin used vries significntly for those EU/EEA countries for which there re dt on consumption. Differences in colistin use might result from mongst others; locl bcteril resistnce sitution, mngement, production type nd vilble mrketing uthoristions. If colistin is no longer vilble then it could be speculted tht other ntimicrobils or mediction (exmple zinc oxide in pig production) would replce its use if no other interventions re tken (biosecurity, vccintion, hygiene ). In recent prospective experimentl study, zinc oxide (ZnO) showed to be s effective s colistin (compred to orl nd in feed groups) on piglet helth nd production prmeters the control of wening dirrhoe, with better dily weight gin during the supplemented period nd reduced dirrhoe score (Vn den Hof et l., submitted). In the cse of zinc oxide, other issues such s environmentl impct nd co-selection of resistnce s for exmple livestock ssocited MRSA should be tken into ccount (Amchwdi et l., 2015; Cvco et l., 2011). The lterntives to colistin, depending on the resistnce sitution in prticulr country, re minopenicillins, trimethoprim-sulphonmides, tetrcyclines, minoglycosides, nd the criticlly importnt ntimicrobil cephlosporins nd fluoroquinolones. The ltter re of prticulr concern due to emerging ESBL resistnce (EFSA BIOHAZ Pnel, 2011; EMEA/CVMP/SAGAM, 2009). Although food-producing nimls re the min concern for the trnsmission of ntimicrobil resistnce from nimls to mn, the risk of trnsmission of ntimicrobil resistnce vi direct contct from compnion nimls should be tken into ccount. Until recently there ws no evidence tht the use of colistin in veterinry medicine for food-producing species hs resulted in the trnsfer of colistin resistnce from nimls to humns. Nevertheless, bsed on current dt, trnsmission of such resistnce is likely to hve tken plce in the EU/EEA, lbeit t low frequency, with the exception of specific cohorts from Asin origin. The results from Chin (Liu et l., 2015; Shen et l., 2016) indicte tht rpid increse cnnot be excluded (Skov nd Monnet, 2016). For other drug resistnt orgnisms including E. coli, the emergence following ntimicrobil consumption nd the trnsfer vi direct niml contct or vi food hs lredy been documented (Angulo et l., 2004). The incresing use of colistin in humns, in prticulr in well-defined settings will led to incresed selection pressure which my be the ctlyst for dispersl of zoonotic colistin Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 26/56

27 resistnce medited by mcr-1 (Skov nd Monnet, 2016). Multifctoril cycling of these reservoirs of genes vi hotspots of colistin use in e.g. intensive cre medicine (Ingenbleek et l., 2015), vi the environment t lrge (Zurfuh et l., 2016) nd fttening poultry, pigs nd vel clves (Cllens et l., 2016) need to be considered in the nlysis of the epidemiology nd for trgeted interventions. The mcr-1 gene hs been found in clinicl cses of veterinry colibcillosis in vel clves nd pigs (Henni et l., 2016; Mlhotr-Kumr et l., 2016; Richez nd Burch, 2016) nd in humn invsive pthogens (Skov nd Monnet, 2016). The mcr-1 genes were found in similr plsmids in the sme bcteri species isolted from food-producing nimls, food humns nd environment indicting possible trnsmission between these comprtments. Dt from 2012 compred fter controlling for biomss in joint report from the Europen Centre for Disese Prevention nd Control (ECDC), the Europen Food Sfety Authority (EFSA) nd EMA, hs shown tht consumption of polymyxins, minly colistin, ws on verge more thn 600 times higher in food-producing nimls thn in humns for the included 19 MSs in the EU nd EEA (ECDC/EFSA/EMA, 2015; Olitn et l., 2015). Since mcr-1 is substntilly more sprse in humns compred to niml isoltes (Kluytmns vn den Bergh et l., 2016) the hypothesis tht it might hve originted from nimls nd then ttin humns is plusible (Skov & Monnet, 2016). The firly low presence in humns so fr, might be due to bsence of selection in non-fvourble environment s indicted by the fct tht ll trvellers tht were tested positive for mcr-1 upon return were negtive fter one month (Arcill et l., 2016). According to Skov & Monnet, the presence of plsmid-medited colistin resistnce in foods nd symptomtic humn crriers combined with incresing colistin use in EU/EEA hospitls my be gme chnger nd the EU/EEA my fce hospitl outbreks of infections with colistin resistnt MDR (Skov & Monnet, 2016). 7. Conclusions on updted literture review Despite its high toxicity, colistin is lst resort ntimicrobil for the tretment of severe infections cused by highly resistnt bcteri in humn medicine (mong others crbpenemse-producing A. bumnnii, P. eruginos, K. pneumonie nd E. coli). Polymyxins with more fvourble toxicologicl profile deserve ttention for further reserch. Following its discovery of the horizontlly trnsferble colistin gene (mcr-1) in 2015, the number of reports is very rpidly incresing with recent increse in niml sources lthough the reltive proportion mid humn clinicl isoltes in the EU/EEA remins firly low (less thn 1%), so fr. Despite the recent nture of the mcr-1 gene discovery, this is n indiction of limited spred of colistin resistnce from food-producing nimls to humn ptients, nd to lesser extent vice vers. The mcr-1 genes were found in similr plsmids in the sme bcteri species isolted from food-producing nimls, food, humns nd environment indicting possible trnsmission between these comprtments. Trnsfer of resistnce either on mobile genetic elements (such s plsmids) between bcteri or from nimls to humns hs been suggested bsed upon prevlence studies but ppers to remin t n overll low incidence in the EU/EEA. It is of therpeutic importnce for the tretment of Grm-negtive gstrointestinl infections in certin food-producing species. From the dt vilble from 26 EU/EEA countries, colistin is the 5th most used ntimicrobil for food-producing nimls (6.1%). There is lrge vrition between MSs in the extent of use of Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 27/56

28 colistin. From the dt vilble the vrition cnnot be directly linked to specific niml species, ctegory or husbndry system in n individul MS with some MSs hving low level, or no use of the substnce, suggesting tht there is scope to decrese the overll use of colistin within the EU. Acquired resistnce mechnisms re no longer limited to stepwise process vi muttions in trget bcteri nd plsmid medited spred is emerging. In humns the clonl resistnce (muttions) forms cn develop rpidly nd cn spred efficiently under certin conditions in hospitls. Since resistnce to other ntimicrobil clsses re frequently found in the sme bcteri tht hrbour mcr-1, this form cn esily spred due to both the use of colistin nd co-selection of other ntibiotic clsses. The mechnisms nd evolutionry pthwys resulting in decresed susceptibility for colistin in certin slmonell serovrs remin to be fully understood. 8. Profiling of the risk to public helth resulting from the use of colistin in nimls in the EU Due to the mjor dt gps relting to risk fctors, prticulrly in reltion to lck of informtion bout the historicl nd current prevlence of colistin resistnce nd the mcr-1 gene, nd its evolution in bcteri in nimls, humns nd food, this risk profiling is bsed substntilly on expert opinion. As new evidence becomes vilble, this profiling my need to be revised Hzrd identifiction Use of colistin in nimls cn select for colistin-resistnt Enterobctericee which hve the potentil to be trnsmitted to humns. In ddition to chromosoml mechnisms of resistnce to colistin, plsmid-borne mechnism hs recently been identified (MCR-1). The mcr-1 gene is ssocited with trnsposble elements locted on different types of plsmids (Kuo et l., 2016; Skov nd Monnet, 2016) nd hs been shown to be present in strins tht hrbour genes encoding for ESBLs nd crbpenemses nd for resistnce to mny other ntimicrobil clsses (Kuo et l., 2016; Poirel et l., 2016). Therefore the use of other ntimicrobil clsses both in humn nd veterinry medicine could mintin mcr-1 colistin resistnce. The potentil for co-selection is high nd colistin-resistnt orgnisms my lso be multi-drug resistnt Exposure Relese of resistnce genes from nimls treted with colistin Colistin is used extensively in food-producing nimls, especilly s group tretments for pigs, poultry nd vel clves. It is mostly dministered vi the orl route nd hs low biovilbility, even mong experimentlly-infected nimls (Rhoum et l., 2015), so direct exposure of the gstrointestinl microbiot is high. The colistin dose used in the EU is bctericidl limiting the selection of resistnt trget orgnisms (Guyonnet et l., 2010); the impct on commensls is less cler. The trnsfer of mcr- 1 plsmids between commensl Enterobctericee hs been shown to be very high in vitro. This hs yet to be demonstrted in vivo but hs the potentil to led to n increse in the previously stble levels of colistin resistnce. The prevlence of colistin-resistnt Slmonell spp. nd E. coli orgnisms in food-producing nimls ppers to be low overll in mjor species. Bsed on the new mechnism of resistnce including the presence of linked resistnce genes, the overll risk for relese of resistnce genes is now ssessed s potentilly high. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 28/56

29 Exposure of humns to resistnce genes vi bcteri from nimls The consumption of pork nd poultry products in the EU is high (consumption of vel is reltively low). Contmintion of met with Slmonell spp. is low, but s with other foodborne orgnisms, dependent on hygiene nd food type mongst other fctors. Although dt re limited, generl prevlence of colistin resistnce in E. coli nd Slmonell spp. from EU produced met ppers to be low, lthough prevlence in poultry nd turkey should be investigted further bsed upon individul country reports (Itly, Germny, Frnce nd the Netherlnds). Exposure to resistnce genes my occur vi other routes, e.g. direct contct with nimls nd mnure in the environment Consequences to humn helth / hzrd chrcteristion Colistin is n ntimicrobil of lst resort in humn medicine tht is used systemiclly to tret serious infections cused by crbpenem-resistnt bcteri tht re generlly lso multi-drug resistnt. As there re often no lterntive tretments for these ptients, the consequences of colistin-resistnt infections re serious (deth). Across the EU, with cler exceptions in defined res, very low numbers of humn ptients require tretment with colistin ech yer nd prevlence of colistin resistnce is low. In recent yers colistin use hs been incresing rpidly in southern Europen regions s consequence of incresing crbpenem resistnce nd this will increse the selection pressure for colistin resistnce. The prospect of new lterntive ntimicrobil substnces coming forwrd in the ner future is very limited, nd lterntive ntimicrobils (e.g. temocillin) re not vilble cross ll countries in the EU/EAA region Overll risk estimtion/chrcteristion A plsmid-borne mechnism of resistnce to colistin (MCR-1) hs recently been identified in Enterobctericee from food-producing nimls. Colistin is used extensively in pigs, poultry nd vel clves, dministered to groups of nimls predominntly vi the orl route. At present, levels of colistin-resistnce in Enterobctericee from nimls re estimted s low; lthough dt on the prevlence of colistin resistnce, including the mcr-1 gene, nd its progression over time re limited. Tking into ccount the nture of veterinry use of colistin, the chrcteristics of the newly identified mechnism of resistnce nd the opportunity for co-selection (Tble 2-Tble 7 nd Tble 9), suggests tht colistin resistnce hs the potentil to spred rpidly nd to be ssocited with MDR orgnisms which could trnsfer to humns, for exmple vi food, litter, or surfce wter. Colistin is used in humn medicine s n ntimicrobil of lst resort for the tretment of serious MDR infections tht re lso resistnt to crbpenems. The occurrence of crbpenem resistnce, subsequent use of colistin, nd therefore its importnce to humn medicine hve incresed substntilly in regions of southern Europe in recent yers. The prospect of novel lterntive ntimicrobils for tretment of these infections in the ner future is limited. In conclusion, lthough there re limited dt on the evolution of colistin resistnce, the newly identified mechnism hs the potentil for rpid spred nd, coupled with the recent incresing importnce of colistin to humn medicine, this leds to n incresed risk to humn helth from the use of colistin in nimls. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 29/56

30 9. Risk Mngement options 9.1. Recommended risk mngement options for colistin in veterinry medicine The min recommendtion is tht colistin sles for use in nimls should be reduced to the minimum fesible nd tht colistin should be dded to more criticl ctegory (ctegory 2) of the AMEG clssifiction (EMA, 2014) (Tble 8). Ctegory 2 includes those ntimicrobil clsses listed s criticlly importnt ntimicrobils by the WHO for which the risk to public helth from veterinry use is considered only cceptble provided tht specific restrictions re plced on their use. Colistin, fluoroquinolones nd 3rd- nd 4th-genertion cephlosporins should be reserved for such use when there re no effective lterntive ntimicrobils for the respective trget species nd indiction. Tble 8. Clssifiction of ntimicrobil clsses ccording to their probbility of trnsfer of resistnce genes nd resistnt bcteri Antimicrobil clss Mobile genetic elementmedited trnsfer of resistnce Verticl trnsmission of resistnce gene(s) b Coselection of resistnce c Potentil for trnsmission of resistnce through zoonotic nd commensl food-borne bcteri d Evidence of similrity of resistnce: genes / mobile genetic elements / resistnt bcteri e Overll probbility of resistnce trnsfer References Assessment 2013 Polymyxins Low (EMA, 2013) (e.g. colistin) Assessment 2016 Polymyxins High UPDATE (e.g. colistin) 2016 Mobile genetic element-medited trnsfer of resistnce. Defined s resistnce gene tht is trnsmitted by mens of mobile genetic elements (horizontl trnsmission of the gene occurs). Probbility (1 to 3): 1, no gene mobiliztion described; 2, gene is exclusively on the core bcteril chromosome; 3, gene is on mobile genetic element, e.g. plsmid. b Verticl trnsmission of resistnce gene. Defined s the verticl trnsfer of resistnce gene through the prent to the dughter bcteri in successful, highly disseminted resistnt clone of bcteri through bcteril popultion, e.g. E. coli ST131 clone, MRSP CC(71) clone, MRSA ST398 clone. Probbility (1 to 3): 1, no verticl trnsmission of gene described s ssocited with in prticulr successful resistnt clone; 2, gene is exclusively on the core bcteril chromosome in prticulr successful resistnt clone; 3, gene is on mobile genetic element, e.g. plsmid, in prticulr successful resistnt clone. c Co-selection of resistnce. Defined s selection of resistnce which simultneously selects for resistnce to nother ntimicrobil. Probbility (1 to 3): 1, no co-mobiliztion of the gene or risk fctor described; 2, gene is either co-mobilized or risk fctor hs been described; 3, gene is co-mobilized nd risk fctor hs been described. d Trnsmission of resistnce through zoonotic nd commensl food-borne bcteri. Defined s trnsmission of resistnce through food-borne zoonotic pthogens (e.g. Slmonell spp., Cmpylobcter spp., Listeri spp., E. coli VTEC) or trnsmission of resistnce through commensl food-borne bcteri (e.g. E. coli, Enterococcus spp.). Probbility (1 to 3): 1, no trnsmission of resistnce through food-borne zoonotic pthogens or commensl food-borne bcteri; 2, trnsmission of resistnce through food-borne zoonotic pthogens or commensl food-borne bcteri; 3, trnsmission of resistnce through food-borne zoonotic pthogens nd commensl food-borne bcteri. e Evidence of similrity of resistnce: genes/mobile genetic elements/resistnt bcteri. Genes - Defined s similr resistnce gene detected in bcteril isoltes of niml nd humn origin; Mobile genetic elements - Defined s similr resistnce mobile genetic element detected in bcteril isoltes of niml nd humn origin; Resistnt bcteri - Defined s similr bcterium hrboring resistnce gene (either chromosomlly or mobile genetic element-encoded) of niml nd humn origin. Probbility (1 to 3): 1, unknown resistnce similrity; 2, genes or mobile genetic elements or resistnt bcteri similr between nimls nd humns; 3, genes nd mobile genetic elements similr between nimls nd humns; 4, genes nd mobile genetic elements nd resistnt bcteri similr between nimls nd humns. The scoring of the tble bove is bsed on the expert opinion of the members of the Working Group. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 30/56

31 Considertions when proposing risk mngement mesures A blnce should be found between the need to protect public helth nd the potentil impct of risk mngement mesures on niml helth ( One Helth pproch). Colistin is minly used in pigs, poultry, nd vel clves to tret E. coli which cuses serious diseses with potentil for high morbidity nd mortlity. Resistnce to ctegory 1 nd other ntibiotics is common. Alterntives to the use of colistin for tretment of the indicted diseses include other criticlly importnt ntimicrobils nd removl of colistin from the mrket could increse the selection pressure for resistnce to these substnces through incresed use. Becuse of the high potentil for co-selection with other clsses, s well s reducing the use of colistin it is importnt tht there is n overll reduction in the use of ntimicrobils of ll clsses. Eliminting ny prophylctic use will be essentil to chieve significnt reduction of sles of colistin for veterinry use. In December 2014 the CVMP recommended to restrict the indictions for use of colistin to tretment of enteric infections cused by susceptible non-invsive E. coli only, tht ny indictions for prophylctic use should be removed nd the tretment durtion limited to the minimum time necessry for the tretment of the disese nd not exceeding 7 dys. In ddition, it ws recommended to remove horses from the SPCs on the grounds of trget species sfety concerns. Commission Decision (2015)1916 of 16 Mrch 2015 trnslted the CVMP recommendtion into legisltion. In April 2016 the CVMP recommended the withdrwl of the mrketing uthoristions for ll veterinry medicinl products for orl use contining colistin in combintion with other ntimicrobil substnces. As colistin is used in ll the mjor food-producing species, mesures in only one niml species would not provide the expected results in terms of reduction of use. Use of colistin s reported to ESVAC (26 countries) decresed 19% between 2011 nd 2013 in terms of tonnes of colistin sold. Countries with low consumption of colistin should be encourged not to increse such use. Trgets should idelly be estblished by niml species, but s comprble consumption dt per niml species cross the EU re not vilble, this is not possible Recommendtion on trget for use of colistin nd considertions on impct on use of other ntimicrobils In order to reduce the exposure of Enterobctericee in nimls to colistin nd hence the possibility of further selection of colistin-resistnce genes which hve the potentil to be trnsmitted to humns, the use of colistin in mg/pcu should be reduced. Such reduction should be chieved without consequentil increse in the consumption (in mg/pcu) of fluoroquinolones, 3rd- nd 4th-genertion cephlosporins or the overll use of ntimicrobils. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 31/56

32 The consumption of ntimicrobils (mount in mg) cn be compred over countries by djusting for the biomss under exposure (kg livestock), which is expressed by the popultion correction unit (PCU). Use of colistin in the EU/EEA countries vries significntly; some EU countries hve reported high consumption of colistin per kg of biomss produced, whilst others hve reported little or no use. Tking into ccount the current use of colistin, the possible lterntives to its use, impcts on niml helth nd welfre nd the tendency over recent yers to reduced consumption of colistin, it is proposed tht there is trget for MSs to reduce use to mximum of 5 mg colistin/pcu (s reported by ESVAC). Further resoning for the trget is provided under justifiction for the trget. If successfully pplied t n EU level, the bove threshold would result in n overll reduction of pproximtely 65% of the current sles of colistin for veterinry use; this decrese should build upon the decrese of colistin sles for veterinry use lredy seen between 2011 nd For those countries with colistin consumption below 5 mg/pcu, the recommendtion should not result in n increse of the colistin consumption. For those countries with consumption tht is well below the proposed 5 mg/pcu, the trends on colistin consumption should be nlysed cse by cse in the concerned country. In some countries with high pig nd poultry production, e.g. Denmrk (0.5 mg/pcu) nd the Netherlnds (0.9 mg/pcu), the level of consumption of colistin is below 1 mg/pcu. Member Sttes should consider the possibility of setting stricter ntionl trgets therefore, idelly lower level thn 5 mg/pcu of colistin, e.g. below 1 mg/pcu, is desirble. There is insufficient informtion to estblish the fesibility of such mesure in ll countries, nd the impct of those intended reductions on colistin resistnce. The recommended im is to chieve the trget for reduction of sles of colistin (in mg/pcu) in period of three to four yers. Through the EU surveillnce progrmmes, the impct of the mesures should be closely monitored nd ssessed to conclude on their impct on ntimicrobil resistnce, including on the presence of the mcr-1 gene in nimls nd humns, if dt re vilble. The reduction of sles of colistin should not be compensted by increse in the use of other clsses; it should be chieved by other mesures such s improved frming conditions, biosecurity in between production cycles, nd vccintion. Due to the possibility of co-selection, n overll reduction of ll ntimicrobils use should be chieved, especilly for those countries for which the ntimicrobil consumption, expressed s mg/pcu, is very high Further considertions Antimicrobil sles dt re not vilble to ESVAC for Greece nd Mlt. Those MSs would need to strt such collection in order to provide the results of colistin sles in mg/pcu. As indicted bove, in cse circumstnces led to significnt increse (or decrese) in the risk to public helth due to the use of colistin in nimls the recommended mesures should be revised. The levels of resistnce to colistin in humns, nimls nd derived foods nd prevlence of mcr-1 herein should be mesured in order to estblish bseline from which to ssess the impct of the mesures. The use of colistin, fluoroquinolones nd 3rd- nd 4th-genertion cephlosporins nd the resons for use, should be recorded by the prescribing veterinrin nd provided to the uthorities s requested. Member Sttes re encourged to set up systems to request nd nlyse these dt. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 32/56

33 Justifiction for the trget One of the objectives when estblishing the trget ws to ensure tht from the current experience from EU countries with high production of pigs nd poultry, it is possible to produce those nimls with consumption of colistin tht is below the proposed trget. The proposed trget is higher thn the current sles of colistin in some countries with high production of pigs nd poultry (i.e. bove 50% PCU). Although the trget will demnd very importnt reduction in the use of colistin for some high using countries (more thn 80% reduction in the most extreme cse), it should still llow for the tretment of nimls in those cses where colistin would remin the best option. It ws considered if the trget should be reduced to below 5 mg/pcu, but reducing the consumption of colistin in high using countries before they hve hd time to implement compenstory strtegies could result in n increse of use of other criticlly importnt ntimicrobils (e.g. fluoroquinolones), or overll use, which could be counterproductive for public helth Summry of the risk mitigtion recommendtions Colistin should be dded to ctegory 2 of the AMEG s clssifiction; the risk to public helth from veterinry use is considered only cceptble provided tht specific restrictions re plced on its use. Colistin, fluoroquinolones nd 3rd- nd 4th-genertion cephlosporins should be reserved for those occsions when there re no effective lterntive ntimicrobils uthorised for the respective trget species nd indiction. There re wide vritions in the use of colistin between countries which re lrgely unexplined. Countries with intensive livestock production cn hve level of usge below 1 mg/pcu (e.g. Denmrk nd the UK) nd much higher, up to mg/pcu (Itly nd Spin). Considering the rpidly incresing importnce of colistin for tretment of criticlly ill humn ptients, ll countries should strive to reduce the use of polymyxins s much s possible. For the current "high nd moderte consumers" the trget nd desirble levels re set t 5 nd 1 or below 1, mg/pcu, respectively, bsed on the observtions on the level of use in other countries. Menwhile more informtion should be gthered to determine the minimum level of colistin use tht cn be chieved while mintining niml welfre nd preventing the incresed use of other criticlly importnt ntimicrobils. If the sitution regrding colistin resistnce in nimls or humns deteriortes further it my be necessry to lower the level proposed trgets. Reduction in use of colistin should be chieved without n increse in the use (in mg/pcu) of fluoroquinolones, 3rd- nd 4th-genertion cephlosporins or overll consumption of ntimicrobils. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 33/56

34 Figure 4. Sles of colistin in for use in nimls in mg/pcu in 2013 (ESVAC dt), including the 5 nd 1 mg/pcu levels. No sles reported in Finlnd, Icelnd nd Norwy Strtegies for responsible use nd lterntives to the use of colistin Strtegies for the responsible use of colistin in veterinry medicine, cn be subdivided into pproches tht limit or fine-tune the use, nd pproches tht replce the use of the substnce. To limit or fine-tune use, better identifiction of nimls tht re disesed versus nimls tht do not need tretment is required. Approprite dignostics should be undertken to estblish the cuse of disese nd identify the pproprite ntimicrobil tretment for the group, if needed. Secondly improving the ntibiotic regimen by pplying PK/PD nlyses to ssist in dose regimen selection (Guyonnet et l., 2010), long with identifying minimum number of dys under exposure is nother option. In recent systemtic review (Burow et l., 2014) it ws concluded tht orlly dministered ntimicrobils increse the risk of ntimicrobil resistnce in E. coli from swine, lthough it ws noted tht more reserch is needed into the impct of dosge nd the longitudinl effects of tretment. Further improved herd mngement, in prticulr biosecurity through well controlled clening nd disinfection strtegies (biocides) (Crlsson et l., 2009), in between production cycles should be encourged to limit the ccumultion of resistnce genes over consecutive production cycles (Dordo- Grcí et l., 2015; Geenen et l., 2011; Schmithusen et l., 2015). Good frming prctices nd herd helth plnning including niml qurntine, restrictions on movements before freedom of disese certifiction, mong others, prevent spred of infections nd therefore reduce the need for ntimicrobils (EFSA/EMA, foreseen 2016). Vccintion, voluntry nd lter mndtory, hs been proved in broilers to reduce the occurrence of Slmonell spp. nd thereby the need for ntimicrobil Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 34/56

35 consumption (Dewele et l., 2012). Vccines re vilble in the EU to reduce the incidence of enteric E. coli infections in piglets. Encourgement should be given to updting vccine ntigen content t regulr intervls to reflect circulting strins. Pro- nd prebiotics, nd in broder sense fecl trnsplnts hve shown in humn medicine to be extremely useful for the control of ntibiotic ssocited dirrhoe (Clostridium difficile) (Cmmrot et l., 2015). Insted of giving long-term doses of ntibiotics vi feed or wter, the digestive trct content cn be replced with helthy bcteri. Given the high number of indictions for ntimicrobils relted to the digestive trct in pigs (Stege et l., 2003), vel clves (Prdon et l., 2012) nd broilers (Persoons et l., 2012), this pproch must receive considertion for further reserch. These historiclly nmed trnsfuntions hve been used for gstrointestinl disorders in horses. Orgnic cids nd metls (Cu, Zn) re lterntives to reduce the use of ntimicrobils t lrge nd colistin in prticulr lthough ttention should be pid to environmentl concerns relting to the use of metls. For n exhustive review on lterntives to replce or to reduce the selection pressure exerted by ntimicrobils in niml husbndry, we refer to the RONAFA working group (Reduction of Need for Antimicrobils in Food-producing Animls) document, to be completed by the end of 2016 (EFSA/EMA, foreseen 2016) Previously pplied risk mngement options Following the previous AMEG recommendtions in 2013, the SPCs for uthorised products were reviewed to ensure consistency for mesures to ensure responsible use in regrds to protecting niml helth nd limiting the possibility of future risk to public helth. As detiled in Section 3.2. referrl ws concluded under Article 35 of Directive 2001/82/EC for ll VMPs contining colistin s sole substnce dministered orlly (including premixes) to food-producing nimls (EMA/CVMP, 2015). Indictions were restricted to therpy or metphylxis, ll indictions for prophylctic use removed nd indictions restricted to the tretment of enteric infections cused by susceptible non-invsive E. coli only. The tretment durtion ws limited to the minimum time necessry for the tretment of the disese nd not exceeding 7 dys. Horses were removed from the SPCs on the grounds of trget species sfety concerns. In April 2016 the CVMP recommended the withdrwl of the mrketing uthoristions for ll veterinry medicinl products for orl use contining colistin in combintion with other ntimicrobil substnces New indictions, formultions or species New indictions, formultions or species (e.g. fish) should be subject to full ntimicrobil resistnce risk ssessment before pprovl. This is the stndrd procedure for ny mrketing uthoristion ppliction for n ntimicrobil product for use in food-producing nimls, but in this cse it is especilly importnt tht the relevnce of colistin for humn medicine is considered for ny new mrketing uthoristion. Studies tht further exmine the effect of different formultions of colistin (polymyxins) on durtion of symptoms, nd excretion of relevnt bcteri nd their ntimicrobil susceptibilities would help to identify nd to decrese inpproprite use. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 35/56

36 9.5. Surveillnce of colistin consumption nd of colistin resistnce The use of colistin in MSs is monitored s prt of the ESVAC project in terms of overll use. The monitoring system should be enhnced to provide figures on use per species, production type nd weight clss. The revised EU/EEA hrmonised monitoring of ntimicrobil resistnce now requires ll MSs to perform stndrdised nd qulity controlled susceptibility testing of colistin on representtive smples of zoonotic nd indictor bcteri (Slmonell spp. nd E. coli). The findings from such testing re reported by MSs s phenotypic dt on colistin resistnce. This monitoring system could be enhnced by selecting rndom smple of resistnt isoltes tht re subsequently screened for resistnce mechnisms, this would fcilitte in prticulr the detection of emerging resistnce genes. Surveillnce of trget niml pthogens isolted from clinicl cses should be implemented to ensure n erly detection of ny chnge on resistnce ptterns. As there is no officil surveillnce of trget niml pthogens, therefore such system should be implemented. The prcticl chllenges for surveillnce re recognised nd re not restricted to colistin Generl considertions Tretment of individul nimls is preferred. Rpid, relible dignostic tests combining ccurte bcteril identifiction (e.g. mss spectrometry) nd colistin susceptibility testing (Liu et l., 2016) should be explored nd tested under routine lbortory conditions. The rpid ccumultion of considerble mount of dditionl informtion following the first report of mcr-1 in November 2015, together with insights in muttions responsible for decresed colistin susceptibility (Wright et l., 2016) highlights the strength of whole genome sequencing (WGS) nd publicly-vilble sequence dtbses (Skov nd Monnet, 2016). Biosecurity mesures, in prticulr in between production cycles, should be implemented to reduce the need for use of ntimicrobils in generl (including colistin) Follow up of the dvice This recommendtion should be reviewed fter three to four yers to determine (i) if the trgets on ntimicrobil consumption hve been chieved, (ii) if possible, if there hs been ny impct on the prevlence of colistin resistnce in food-producing nimls, lthough cknowledging tht there re limited dt, especilly in regrds to the mcr-1 gene nd tht more time might be required to observe chnges in resistnce levels. At this time, further considertion should be given to ny chnges in the need for nd use of colistin in humn medicine nd the occurrence of colistin resistnce in humns. The effectiveness of the proposed mesures should then be reviewed tking One Helth pproch, nd further considertions on the mesures s detiled in the Annex (section 1) should be ddressed. Further studies on the mechnism nd routes of trnsmission of colistin resistnce from nimls to humns would be useful to clrify the res where informtion vilble is limited. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 36/56

37 ANNEX 1. Risk Mngement options tht were nlysed nd disregrded 1.1. Withdrwl of existing mrketing uthoristions The withdrwl of mrketing uthoristions ws considered but it ws noted tht, in ddition to potentil niml helth nd welfre impcts, this could increse the use of other CIAs, in prticulr fluoroquinolones, s there re high levels of resistnce to ctegory 1 nd other lterntive substnces for the given indictions. It could be speculted tht due to the high potentil for co-selection by other ntimicrobil clsses, the mcr-1 gene would still be mintined in niml popultions fter withdrwl of colistin Group tretments The option of plcing restrictions to reduce the use of colistin for the tretment of groups of nimls ws discussed. Approximtely 99% of use of colistin is in orl formultions which re mostly used for group tretment within herds/flocks. The sme resons s provided bove for not recommending the withdrwl of existing mrketing uthoristions pply for not bnning group tretment. It ws lso considered if premix formultion should be withdrwn since these could hve greter tendency to be used off-lbel for prolonged durtion of (preventive) tretment. ESVAC dt suggest tht in those MSs where use of medicted feeds is limited, this does not necessrily impct colistin sles nd orl powder nd solution formultions re used insted. In ddition, due to differences in use of premix nd other orl formultions tht my be ssocited with vilbility nd ntionl legisltion, this mesure would be inconsistent cross the EU Restriction on use for metphylxis As 99% of use of colistin is in orl formultions which re mostly used for simultneous group tretment nd metphylxis within herds/flocks, nd it is difficult to seprte mediction of cliniclly ill nd in-contct nimls in intensive husbndry systems, it ws considered tht this mesure would not be prcticl to implement effectively Restriction from use in certin species Sufficient species-specific dt re not vilble to perform the risk ssessment required Injectble, intrmmmry nd topicl formultions Tking into ccount the fct tht these formultions ccount for less thn 1% of colistin sles, re mostly used for individul niml tretment nd vi non-enterl routes of dministrtion, it ws considered tht lthough colistin should be in Ctegory 2, further restrictions on the use of these colistin formultions would not hve mjor impct on the risk to public helth. If in the future it is pprent the sles of these formultions re incresing, then the possibility of the restrictions of the use should be reconsidered. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 37/56

38 2. Figures Figure 5. Sptil distribution of sles of polymyxins in veterinry medicine, in mg/kg biomss, in 26 EU/EEA countries, for No sles reported in Finlnd, Icelnd nd Norwy. (EMA/ESVAC, 2015) Figure 6. Sptil distribution of sles of polymyxins in humn medicine, in mg/kg biomss, in 25 EU/EEA countries, for 2013 (dt shown only for countries reporting on totl consumption in the country; i.e. reporting for ntibiotic consumption in the community (outside hospitls) nd in the hospitl sector) (ECDC, 2015) Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 38/56

39 Plese note tht Figure 5 nd Figure 6 show polymyxin consumption expressed in mg/kg biomss with different scle becuse consumption is much lower in humns thn in nimls. Figure 7. Percentge of veterinry sles in mg/pcu for food-producing nimls, by phrmceuticl form of polymyxins, in the EU/EEA for No sles reported in Finlnd, Icelnd nd Norwy (EMA/ESVAC, 2015) (unpublished ESVAC dt 2013) *Negligible mount of polymyxins were sold s orl pste, bolus, intrmmmry nd intruterine preprtions. Figure 8. Copy of the Februry 2016 cll for scientific dt for the updte of dvice Advice on the impct on public helth nd niml helth of the use of ntibiotics in nimls (colistin) following the recent discovery of the first mobile colistin resistnce gene (mcr-1) Cll for scientific dt for the updte of dvice Submission period: 29 Februry 15 Mrch 2016 Der collegues, The CVMP nd CHMP invites ll interested prties to submit ny scientific dt which might hve impct on public nd niml helth tht should be considered when updting the previously published dvice on colistin. The nswers should ddress some of the following points: The importnce of colistin to humn nd veterinry medicine (e.g. estimted frequency of use, trget indictions, including selective digestive trct decontmintion, estimtion of the use per niml species). Any informtion on colistin resistnce medited by the mcr-1 gene in isoltes from humns nd nimls, including niml pthogens. The effectiveness nd vilbility of lterntive tretments to the use of colistin in humn nd nimls especilly if restrictions on the use of colistin would be pplied. Updted dvice on the use of colistin products in nimls within the Europen Union: development of resistnce nd possible impct on humn nd niml helth EMA/CVMP/CHMP/231573/2016 Pge 39/56