American Thoracic Society DocumentsAn Official ATS/IDSA Statement: Diagnosis,Treatment, and Prevention of NontuberculousMycobacterial DiseasesDavidE. Griffith, Timothy Aksamit, BarbaraA. Brown-Elliott, AntoninoCatanzaro, Charles Daley, FredGordin,StevenM. Holland, Robert Horsburgh, GwenHuitt, Michael F. Iademarco, Michael Iseman, KennethOlivier,StephenRuoss, C. FordhamvonReyn, RichardJ. Wallace, Jr., andKevinWinthrop, onbehalf of theATSMycobacterial Diseases SubcommitteeThis Ofcial Statement of the American Thoracic Society (ATS) and the Infectious Diseases Society of America (IDSA)was adopted by the ATS Board Of Directors, September 2006, and by the IDSA Board of Directors, January 2007CONTENTSSummaryDiagnostic Criteria of Nontuberculous MycobacterialLung DiseaseKey Laboratory Features of NTMHealth Care- and Hygiene-associatedDisease PreventionProphylaxis and Treatment of NTM DiseaseIntroductionMethodsTaxonomyEpidemiologyPathogenesisHost Defense and Immune DefectsPulmonary DiseaseBody MorphotypeTumor Necrosis Factor InhibitionLaboratory ProceduresCollection, Digestion, Decontamination, and Stainingof SpecimensRespiratory SpecimensBody Fluids, Abscesses, and TissuesBloodSpecimen ProcessingSmear MicroscopyCulture TechniquesIncubation of NTM CulturesNTM IdenticationAntimicrobial Susceptibility Testing for NTMMolecular Typing Methods of NTMClinical Presentations and Diagnostic CriteriaPulmonary DiseaseCystic FibrosisHypersensitivity-like DiseaseTransplant RecipientsDisseminated DiseaseLymphatic DiseaseSkin, Soft Tissue, and Bone DiseaseThisdocumenthasanonlinesupplement,whichisaccessiblefromthisissuestable of contents at www.atsjournals.orgAm J Respir Crit Care Med Vol 175. pp 367416, 2007DOI: 10.1164/rccm.200604-571STInternet address: www.atsjournals.orgHealth Care and Hygiene-associated Disease and DiseasePreventionNTM Species: Clinical Aspects and Treatment GuidelinesM. avium Complex (MAC)M. kansasiiM. abscessusM. chelonaeM. fortuitumM. genavenseM. gordonaeM. haemophilumM. immunogenumM. malmoenseM. marinumM. mucogenicumM. nonchromogenicumM. scrofulaceumM. simiaeM. smegmatisM. szulgaiM. terrae complexM. ulceransM. xenopiOther NTM Species/PathogensResearch Agenda for NTM DiseaseSummary of RecommendationsSUMMARYDiagnostic Criteria of Nontuberculous MycobacterialLung DiseaseThe minimum evaluation of a patient suspected of nontubercu-lous mycobacterial (NTM) lung disease should include the fol-lowing: (1)chest radiographor, intheabsenceof cavitation,chest high-resolution computed tomography (HRCT) scan; (2)threeor moresputumspecimens for acid-fast bacilli (AFB) analysis;and(3)exclusionofotherdisorders, suchastuberculosis(TB).Clinical, radiographic, and microbiologic criteria are equally im-portant and all must be met to make a diagnosis of NTM lungdisease. Thefollowingcriteriaapplytosymptomaticpatientswith radiographic opacities, nodular or cavitary, or an HRCTscanthat shows multifocal bronchiectasis with multiple small nodules.These criteria t best with Mycobacteriumaviumcomplex (MAC),M. kansasii, and M. abscessus. There is not enough known aboutmost other NTM to be certain that these diagnostic criteria areuniversally applicable for all NTM respiratory pathogens.368 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 2007Clinical.1. Pulmonarysymptoms, nodular or cavitaryopacities onchest radiograph, or an HRCT scan that shows multifocalbronchiectasis with multiple small nodules.and2. Appropriate exclusion of other diagnoses.Microbiologic.1. Positive culture results from at least two separate expecto-rated sputum samples. (If the results from the initial spu-tumsamplesarenondiagnostic, considerrepeatsputumAFB smears and cultures.)or2. Positiveculture results from at leastone bronchial washor lavage.or3. Transbronchial orotherlungbiopsywithmycobacterialhistopathologic features (granulomatous inammation orAFB)andpositivecultureforNTMorbiopsyshowingmycobacterial histopathologic features (granulomatous in-ammation or AFB) and one or more sputum or bronchialwashings that are culture positive for NTM.4. ExpertconsultationshouldbeobtainedwhenNTMarerecovered that are either infrequently encountered or thatusually represent environmental contamination.5. Patients who are suspected of having NTM lung disease butwho do not meet the diagnostic criteria should be followeduntil the diagnosis is rmly established or excluded.6. Making the diagnosis of NTM lung disease does not, perse, necessitate the institution of therapy, which is a decisionbased on potential risks and benets of therapy for individ-ual patients.Key Laboratory Features of NTM1. The preferred staining procedure is the uorochromemethod. Specimens shouldbeculturedonbothliquidand solid media. Species that require special growth con-ditions and/or lower incubation temperatures includeM. haemophilum, M. genavense, and M. conspicuum. Thesespecies can cause cutaneous and lymph node disease.2. In general, NTM should be identied to the species level.Methods of rapid species identication include commercialDNA probes (MAC, M. kansasii, and M. gordonae) andhigh-performance liquidchromatography (HPLC). Forsome NTM isolates, especially rapidly growing mycobac-terial (RGM) isolates (M. fortuitum, Mabscessus, andM. chelonae), other identication techniques may be nec-essary including extended antibiotic in vitro susceptibilitytesting, DNAsequencingor polymerasechainreaction(PCR) restriction endonuclease assay (PRA).3. RoutinesusceptibilitytestingofMACisolatesisrecom-mended for clarithromycin only.4. Routine susceptibility testing of M. kansasii isolates is rec-ommended for rifampin only.5. Routine susceptibility testing, for both taxonomic identi-cation and treatment of RGM (M. fortuitum, M abscessus,and M. chelonae) should be with amikacin, imipenem(M. fortuitumonly), doxycycline, the uorinated quinolones,a sulfonamide or trimethoprim-sulfamethoxazole, cefoxitin,clarithromycin, linezolid, and tobramycin (M. chelonaeonly).Health Care and Hygiene-associated Disease PreventionPrevention of health carerelated NTM infections requires thatsurgicalwounds, injectionsites,andintravenouscathetersnotbe exposed to tap water or tap waterderived uids. Endoscopescleaned in tap water and clinical specimens contaminated withtap water or ice are also not acceptable.Prophylaxis and Treatment of NTM Disease1. Treatment of MAC pulmonary disease. For most patientswith nodular/bronchiectatic disease, a three-times-weeklyregimen of clarithromycin (1,000 mg) or azithromycin (500mg), rifampin(600mg), andethambutol (25mg/kg) isrecommended. For patients with brocavitary MAC lungdiseaseorseverenodular/bronchiectaticdisease, adailyregimen of clarithromycin (5001,000 mg) or azithromycin(250mg), rifampin(600mg)orrifabutin(150300mg),andethambutol(15mg/kg)withconsiderationofthree-times-weekly amikacin or streptomycin early in therapy isrecommended. Patients shouldbetreateduntil culturenegative on therapy for 1 year.2. Treatment of disseminated MAC disease. Therapy shouldinclude clarithromycin (1,000 mg/d) or azithromycin(250 mg/d) and ethambutol (15 mg/kg/d) with or withoutrifabutin(150350mg/d). Therapycanbediscontinuedwithresolutionof symptomsandreconstitutionof cell-mediated immune function.3. Prophylaxis of disseminatedMACdisease. Prophylaxisshould be given to adults with acquired immunodeciencysyndrome(AIDS)withCD4T-lymphocytecountslessthan 50 cells/l. Azithromycin 1,200 mg/week or clarithro-mycin 1,000 mg/day have proven efcacy. Rifabutin300 mg/day is also effective but less well tolerated.4. Treatment of M. kansasii pulmonary disease. A regimenof dailyisoniazid(300mg/d), rifampin(600mg/d), andethambutol (15 mg/kg/d). Patients should be treated untilculture negative on therapy for 1 year.5. Treatment of M. abscessus pulmonary disease. There arenodrugregimensof provenorpredictableefcacyfortreatment of M. abscessuslungdisease. Multidrugregi-mens that include clarithromycin 1,000 mg/day may causesymptomatic improvement and disease regression. Surgi-cal resection of localized disease combined with multidrugclarithromycin-basedtherapyoffersthebest chanceforcure of this disease.6. Treatment of nonpulmonary disease caused by RGM(M. abscessus, M. chelonae, M. fortuitum). The treatmentregimen for these organisms is based on invitro susceptibil-ities. For M. abscessus disease, a macrolide-based regimenis frequentlyused. Surgicaldebridement mayalsobe animportant element of successful therapy.7. Treatment of NTM cervical lymphadenitis. NTM cervicallymphadenitis is due to MAC in the majority of cases andtreated primarily by surgical excision, with a greater than90% cure rate. A macrolide-based regimen should be con-sidered for patients with extensive MAC lymphadenitis orpoor response to surgical therapy.INTRODUCTIONThis is the third statement in the last 15 years dedicated entirelytodiseasecausedbyNTM(1, 2). Thecurrentunprecedentedhigh level of interest in NTM disease is the result of two majorrecent trends: theassociationof NTMinfectionwithAIDSandrecognitionthat NTMlungdiseaseis encounteredwithAmerican Thoracic Society Documents 369increasing frequency in the non-AIDS population. Furthermore,NTMinfectionsareemerginginpreviouslyunrecognizedset-tings, withnewclinical manifestations. Another majorfactorcontributing to increased awareness of the importance of NTMas human pathogens is improvement in methodology in the my-cobacteriologylaboratory, resultinginenhancedisolationandmorerapidandaccurateidenticationof NTMfromclinicalspecimens. Consistent withthe advances in the mycobacteriologylaboratory, this statement has an emphasis on individual NTMspecies and the clinical diseasespecic syndromes they produce.A major goal is facilitating the analysis of NTM isolates by thehealth care provider, including determination of the clinical andprognostic signicance of NTM isolates and therapeutic options.There are controversies in essentially all aspects of this verybroad eld and, whenever possible, these controversies are high-lighted. Hence, an attempt is made to provide enough informa-tionsothattheclinician understandsthe recommendationsintheir appropriate context, especially those made with inadequateorimperfectsupportinginformation. Also, whenthereisnotcompellingevidenceforone recommendation, alternativerec-ommendations or options are presented.This statement alsoincludes newtopics not addressedinprevious statements, including advances in the understanding ofthe pathogenesis of NTMdisease, descriptions of new NTM patho-gens, clinical areas of emerging NTM disease such as cystic bro-sis, newNTMdiseasemanifestationssuchashypersensitivity-like lung disease, and public health implications of NTM diseasesuchaspreventionandsurveillance. Thisstatement will alsotakeadvantage ofweb-based resourcesthrough the AmericanThoracic Society (ATS) website for illustration and amplicationofselectedtopics, discussionofadditionaltopicsnotincludedin this document, as well as the capacity for updating informationin this rapidly changing eld.Large gaps still exist in our knowledge. Limitations in system-atic data have made it necessary for many of the recommenda-tions in this document to be based on expert opinion rather thanonempiricallyderived data.Unquestionably,moreandbetterstudies of NTMdiseases must be done. The committee organizedto create this document is composed of scientists and physiciansresiding in the United States. This document, therefore, repre-sentsaUnitedStates perspectiveofNTMdiseasesthatmaynot be appropriate for NTM diseases in other parts of the world.METHODSReviewof theavailableliteratureandsubsequent gradingofrecommendationswereaccomplishedbythemembersof thewriting committee. The searchfor evidence includedhand-searching journals, reviewing previous guidelines, and searchingelectronicdatabasesincludingMEDLINEandPubMed. Onlyarticles written in English were considered. Final decisions forformulating recommendations were made by voting among com-mittee members. The recommendations are rated on the basisofasystemdevelopedbytheU.S. PublicHealthServiceandtheInfectious DiseasesSocietyof America(IDSA) (3). Theratingsystemincludes aletter indicatingthestrengthof therecommendation, andaromannumeral indicatingthequalityof the evidence supporting the recommendation (3) (Table 1).Thus, clinicians can use the rating to differentiate among recom-mendationsbasedondata from clinicaltrialsandthose basedon the opinions of the experts comprising the writing committee,who are familiar with the relevant clinical practice and scienticrationale for suchpractice when clinical trial data are not available.Ratings following each numbered recommendation pertain to allpointswithin thenumbered recommendation. Each memberofthe writing committee has declared any conict of interest. Theseguidelines were developed with funding provided by the ATS.TAXONOMYWhen the last ATS statement about NTM was prepared in 1997,there were approximately 50 NTM species that had been identi-ed. Currently, more than 125 NTMspecies have been cataloged(4,5). AlistofNTM speciesidentied since1990 isprovidedin the online supplement. A comprehensive list of all validatedNTMspecies canbefoundonlineat www.bacterio.cict.fr/m/mycobacterium.html. There has been a dramatic recent increasenot only in the total number of mycobacterial species but alsoin the number of clinically signicant species. Clinicians mightreasonablyask,WhyaretheresomanynewNTMspecies?Theincrease relatestoimproved microbiologic techniques forisolating NTM from clinicalspecimens and, more importantly,to advances in molecular techniques with the development andacceptanceof 16SrRNAgenesequencingas astandardfordening new species.Earlytaxonomicstudies comparedupto100growthandbiochemical tests of large numbers of strains in multiple collabo-rative laboratories. Work focused around the InternationalWorking Group on Mycobacterial Taxonomy. New species weredened on the ability to phenotypically separate the new taxonfrom established species. This work was time and labor intensiveandnot adequatefor separatingmanyNTMspecies. Subse-quently, species were identiedby comparisons of genomicDNA; new species had similarity (homology) of less than 70%onDNADNApairingexperimentswithestablishedspecies.This type of comparison was highly technical, highly labor inten-sive, andrequiredcomparisonof possiblenewspecies toallestablishedrelatedspecies. Byitsverynature, thistechniquelimited identication of new species.The dramatic change in mycobacterial taxonomy came withthe ready availability and reliability of DNAsequencing. Investi-gatorsrecognizedthatthemycobacterial16SrRNAgenewashighly conserved, and that differences in the sequence of 1% orgreatergenerallydenedanewspecies(4, 5).Alsocriticaltothedramaticchangewastheappearanceofpubliclyavailabledatabasesthat storedthe 16S rRNA genesequences ofestab-lished mycobacterial species. Recognition of a novel NTM spe-cies is now relatively simple: perform 16S rRNA gene sequenceanalysis of a suspected new species and compare the results withthoseinthedatabases. Numerousnewspeciesareappearingfromlaboratoriesallovertheworldratherthanfromasmallnumber of mycobacterial taxonomists. It is likely that the numberofnewspecieswill continuetoexpandrapidlyas16SrRNAgenesequencinganalysisisperformedonincreasingnumbersof isolates of clinical disease that cannot be identied with com-mercial nucleicacidprobes. ItispossiblethatthenumberofNTM species will increase to more than 150 before the publica-tion of the next NTM disease statement.Thisexpansionin new NTMspecies is, therefore,largelyaconsequence of newer identication techniques that are capableofseparatingcloselyrelatedNTMspecies, asopposedtothesudden appearance of newNTMspecies. For instance, M. triplex,M. lentiavum, M. celatum, and M. conspicuum (among others)might previously have been identied as MAC based on tradi-tional biochemical and/or phenotypic analyses. The clinical sig-nicance of these species separations may be subtle or negligible,buttheclinicianinevitablywill continuetobeconfrontedbynew NTM species designations.EPIDEMIOLOGYNTM are widely distributed in the environment with high isola-tion rates worldwide (6, 7). Organisms can be found in soil and370 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 2007TABLE 1. THESTRENGTHOFRECOMMENDATIONSBASEDONQUALITYOFEVIDENCE(ADAPTEDFROMTHEINFECTIOUSDISEASESOCIETYOFAMERICA/UNITEDSTATESPUBLICHEALTHSERVICERATINGSYSTEM)Categories Reflecting the Strength of Each Grades Reflecting the Quality of Evidence onRecommendation for or against Its Use Which Recommendations Are BasedCategory Definition Grade DefinitionA Good evidence to support a recommendation for use I Evidence from at least one properly randomized, controlled trialB Moderate evidence to support a recommendation II Evidencefromatleastonewell-designedclinical trial withoutrandomization,fromcohortfor use or case-controlled analytic studies (preferably from more than one center), from multipleC Poor evidence to support a recommendation for or time-series studies or from dramatic results in uncontrolled experimentsagainst use III Evidence from opinion of respected authorities, based on clinical experience, descriptiveD Moderate evidence to support a recommendation studies, or reports of expert committeesagainst useE Good evidence to support a recommendation against usewater, includingbothnatural andtreatedwater sources (M.kansasii, M. xenopi, and M. simiae are recovered almost exclu-sivelyfrommunicipal watersourcesandrarely, if ever, fromother environmental sources). When identical methods are used,isolationratesofNTMfromtheenvironmentareremarkablysimilar in diverse geographic areas (6, 7). There is no evidenceof animal-to-human or human-to-human transmission of NTM(711). Even in patients with cystic brosis (CF), an apparentlyhighlysusceptiblepopulation andapopulation inwhich otheropportunistic organisms areclearlypassedbetweenpatients,there has been no documentation of human-to-human transmis-sion of NTM (12). Human disease is suspectedto be acquiredfrom environmental exposures, although the specic source ofinfection usually cannot be identied (13).NTM may cause both asymptomatic infection and symptom-aticdiseaseinhumans. Ratesofasymptomaticinfectionhavebeen inferred from both antibody and skin test studies. In areaswhere infection with M. tuberculosis is uncommon, antibody toacommonmycobacterial antigen, lipoarabinomannin(LAM),canbeassumedtobeduepredominantlytoNTMinfection.Antibody to LAM is detectable among children in the UnitedStates, rising rapidlybetween the ages of 1 and12 years, thenappearingtoplateau(14). Skinteststudiesinadultsindicatethat asubstantial proportionhavehadpriorandpresumablyasymptomaticinfectionwithNTM(15, 16). Skintest studieswithanM. intracellularepuriedproteinderivative(PPD-B)conducted among U.S. Navy recruits in the 1960s showed thatreactions of greater than 4 mm induration were more common inthe southeastern than northern United States, suggesting higherbackground rates of NTMinfectioninthese areas (16). Reactionsto skin tests derived from NTM are not sufciently species spe-cic toindicatewhichnontuberculous mycobacteriummighthave been responsible for these asymptomatic infections, and itispossiblethat cross-reactivitywithM. tuberculosisinfectioncontributed to some of these reactions. However, because MACorganisms are the mostcommon cause of NTM diseasein theUnited States, it is likely that MAC was also the most commoncause of infection. In these patients, asymptomatic infection withNTM has not been shown to lead to latent infection, so that incontrast toTB, thereiscurrentlynoevidencethatNTMareassociated with reactivation disease.NTMdiseaseshavebeenseeninmostindustrializedcoun-tries; incidencerates varyfrom1.0to1.8cases per 100,000persons (17). These overall incidence rates are estimates basedon numbers of NTMisolates reported. MACis the most commonNTMspeciescausingdisease inmostseries,but manyspecieshave beenimplicated(Table 2).Rates appear tobesimilar inmostdevelopedcountries, but surveillanceinformationislim-ited. Because NTM diseases are not communicable, they are notreportable in the United States. Although several reports havesuggestedthat theincidenceof NTMdiseases has increasedoverthepast several decades, thisobservationhas not beenconclusively established due to the lack of comprehensive sur-veillanceefforts. Themost commonclinical manifestationofNTM disease is lung disease, but lymphatic, skin/soft tissue, anddisseminated disease are also important (18, 19). In the UnitedStates between 1981 and 1983, 94%of the NTMisolates reportedto the Centers for Disease Control and Prevention (CDC) labo-ratory were pulmonary, whereas 3% were lymph node and 3%wereskin/softtissueisolates(19). Inthelate1980sandearly1990s, NTMisolates associatedwithdisseminateddiseaseinpatients with AIDSwere reported almost as frequently as pulmo-naryisolates(20, 21). Morerecently, theCDCpublishedtheresults of NTM isolates reported by state public health labora-tories to the Public Health Laboratory System(PHLIS) databasefortheperiod1993through1996(22). Duringthisperiod, ofthe NTM isolated, 75% were pulmonary, whereas 5% were fromblood, 2%fromskin/soft tissue, and0.4%fromlymphnodeisolates. Themost frequentlyreportedpotentiallypathogenicspeciesandcorrespondingreportratesoverthe4-yearperiod(per 1,000,000 population) were as follows: MAC, 29 to 36 iso-lates; M. fortuitum, 4.6to6isolates; andM. kansasii, 2to3.1isolates. Regionally, all three NTM species were reported mostoften from the southeastern United States. There are, however,signicant limitations interpreting and extrapolating these data.Reporting of NTM species suspected to be involved in diseaseto PHLIS was voluntary, and not all states participated. There-fore, thedatadonot represent theabsoluteoccurrenceanddistribution ofNTMspecies in the UnitedStates. Inaddition,thenumbers of isolates arepresentedas report rates,toreect a lack of verication of clinical signicance of the report(emphasis inreport). Becausethereports werenot veried,interpretingtheassociationoftheseisolateswithclinicaldatais problematic.Overall, there is not substantially more or better informationabout NTM disease prevalence than that published in the 1997ATS statement on NTM, except that, in most state public healthlaboratories, NTM isolates, especially MAC isolates, are morecommon than M. tuberculosis isolates (21, 22).Moreepidemiologicinformationisprovidedinthediscus-sions of specic disease syndromes and for individual NTM spe-cies, and in Table 2.PATHOGENESISOver the past two decades, three important observations havebeen made regarding the pathogenesis of NTM infections:American Thoracic Society Documents 371TABLE 2. CLINICALDISEASECAUSEDBYNONTUBERCULOUSMYCOBACTERIA(ALPHABETICALORDERBYSPECIES)Common Page Comment Uncommon Page CommentPulmonary DiseaseM. abscessus 396 Worldwide; may be found concomitant with MAC M. asiaticum* Rarely isolatedM. avium complex 386 Worldwide; most common NTM pathogen in U.S. M. celatum* Cross-reactivity with TB-DNA probeM. kansasii 395 U.S., Europe, South Africa, coal-mining regions M. chelonae 398M. malmoense 399 U.K., northern Europe; uncommon in U.S. M. fortuitum 398 Associated with aspirationM. haemophilum 399 Rarely isolatedM. xenopi 402 Europe, Canada; uncommon in U.S.; associated with M. scrofulaceum 400 South Africa; uncommon in U.S.pseudoinfectionM. shimoidei* Rarely isolatedM. simiae 401 Southwest U.S., associated with pseudo-outbreaksM. smegmatis 401 Rarely isolatedM. szulgai 401 Rarely isolated, not an environmental contaminantLymphadenitisM. avium complex 386 Worldwide; most common NTM pathogen in U.S. M. abscessus Rarely isolatedM. malmoense 399 U.K., northern Europe (especially Scandinavia) M. chelonae 398M. scrofulaceum 400 Worldwide; previously common, now rarely isolated M. fortuitum 398in U.S. M. genavense 399 Fastidious species (See LABORATORY PROCEDURES)M. haemophilum 399 Fastidious species (See LABORATORY PROCEDURES)M. kansasii Rarely isolatedM. szulgai 401 Rarely isolatedDisseminated DiseaseM. avium complex 386 Worldwide; AIDS; most common NTM M. abscessus 396 Non-AIDS immunosuppressedpathogen in U.S. M. celatum* AIDSM. chelonae 398 U.S.; non-AIDS immunosuppressed skin lesions M. conspicuum* AIDS, non-AIDS immunosuppressedM. fortuitum 398 Non-AIDS immunosuppressedM. haemophilum 399 AIDS; U.S., Australia; non-AIDS immunosuppressed M. genavense 399 AIDSM. kansasii 395 AIDS; U.S., South Africa M. immunogenum 399 Rare, associated with pseudo-outbreaksM. malmoense 399 U.K., northern Europe; non-AIDS immunosuppressedM. marinum 400 Worldwide; AIDSM mucogenicum 400 Central venous catheter infectionsM. scrofulaceum 400 Rarely isolatedM. simiae 401 Southwest U.S., associated with pseudoinfectionM. szulgai 401 Rarely isolatedM. xenopi 402 Europe, Canada, associated with pseudoinfectionSkin, Soft Tissue, and Bone DiseaseM. abscessus 396 Penetrating injury M. avium complex 386 WorldwideM. chelonae 398 U.S., associated with keratitis and M. haemophilum 399 Extremities, cooler body sitesdisseminated disease M. immunogenum 399 Rarely isolated, associated with pseudo-outbreaksM. fortuitum 398 Penetrating injury, footbaths M. kansasii Rarely isolatedM. marinum 400 Worldwide, fresh- and saltwater M. malmoense 399 U.K., northern EuropeM. ulcerans 402 Australia, tropics, Africa, Southeast Asia, not U.S. M. nonchromogenicum 400 TenosynovitisM. smegmatis 401 Rarely isolatedM. szulgai 401 Rarely isolatedM. terrae complex 402 TenosynovitisSpecimen ContaminantM. gordonae 399 Most common NTM contaminantM. haemophilum 399M. mucogenicum 400M. nonchromogenicum 400M. terrae complex 402Definition of abbreviations: MAC Mycobacterium avium complex; NTM nontuberculous mycobacteria.* See the online supplement.1. Inpatients infected with HIV, disseminated NTMinfec-tions typically occurred only after the CD4T-lymphocytenumberhadfallenbelow50/l, suggestingthat specicT-cell products or activities are required for mycobacterialresistance (23, 24).2. In the HIV-uninfected patient group, genetic syndromesof disseminated NTM infection have been associated withspecicmutationsininterferon(IFN)-andinterleukin(IL)-12 synthesis and response pathways (25, 26) (IFN-receptor1[IFNR1],IFN-receptor2[IFNR2],IL-12receptor 1subunit [IL12R1], theIL-12subunit p40[IL12p40], the signal transducer and activator of transcrip-tion 1 [STAT1], and the nuclear factor- essential modu-lator [NEMO]).3. There is also an association between bronchiectasis, nodu-lar pulmonary NTM infections and a particular body habi-tus, predominantly in postmenopausal women (e.g., pectusexcavatum, scoliosis, mitral valve prolapse) (27).Host Defense and Immune DefectsMycobacteria are initially phagocytosed by macrophages, whichrespondwithproductionofIL-12, whichinturnup-regulates372 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 2007IFN-(28). IFN-activates neutrophils andmacrophages tokill intracellular pathogens, including mycobacteria. There is apositive feedback loop between IFN- and IL-12, which is criticalfor the control of mycobacteria, as well as certain other intracel-lular infections. Disseminated NTMdisease is a denite manifes-tation of immunologic defect, either acquired, such as HIV andiatrogenicfactors, orgenetic, causedbydefectsintheaboveIFN-/IL-12 pathway genes. However, these genetic factors onlypredispose to disseminated disease.Pulmonary DiseaseLung disease due to NTM occurs commonly in structural lungdisease, such as chronic obstructive pulmonary disease (COPD),bronchiectasis, CF, pneumoconiosis, prior TB, pulmonary alveo-lar proteinosis, and esophageal motility disorders (12, 19, 2932).Abnormal CF genotypes and1-antitrypsin (AAT) phenotypesmay predispose some patients to NTM infection (3335). NTMlungdiseasealsooccursinwomenwithoutclearlyrecognizedpredisposing factors (32, 3638). Bronchiectasis and NTM infec-tion, usuallyMAC, oftencoexist, makingcausalitydifculttodetermine. These patients may carry multiple MAC strains overtime, suggesting either polyclonal infection or recurrent infectionwithdistinct strains(38). It isunclear whetherthisproblem isduetolocalabnormalities(e.g.,bronchiectasis)ortoimmunedefects. In one study from Japan, 170 patients with MAC lunginfectionwerestudied: of622 siblings ofthose patients, 3hadMAClungdisease.TheimplicationisthatthesiblingriskforMACinfection is much higher than previously estimated popula-tion prevalence (11).Body MorphotypeWomen with nodular NTMpulmonary infections associated withbronchiectasis have similar clinical characteristics and body type,sometimesincludingscoliosis, pectusexcavatum, mitral valveprolapse, and joint hypermobility (27). These phenotypic charac-teristics may represent markers for specic genotypes that affectboth body morphotype and NTM infection susceptibility. Alter-natively, the morphotype itself may inuence mycobacterial in-fection susceptibility, through suchfeatures as poor tracheobron-chial secretion drainage or ineffective mucociliary clearance.Tumor Necrosis Factor InhibitionIFN- and IL-12 control mycobacteria in large part through theup-regulation of tumor necrosis factor (TNF)-, made predomi-nantlybymonocytes/macrophages. ThecriticalroleofTNF-in controlling intracellular infections is made clear through theuse of TNF- blocking agents. The potent TNF- blocking anti-bodies iniximab and adalimumab and the soluble receptor etan-erceptareeffectiveantiinammatoryagentsandleadtorela-tively high rates of development of active TB in those who arelatently infected (39, 40). The onset of TB after administrationof iniximab ranges fromweeks to months. In addition to TB, theTNF- blocking agents predispose to invasive fungal infections,such as aspergillus, histoplasmosis, and coccidioidomycosis (41).Infections with mycobacteria and fungi are seen with all threeagents, butsignicantlymorewithiniximabthanetanercept.Adalimumabshouldberegardedashavingsimilarrisks. Therisk posed by TNF- blocking agents for predisposing to NTMinfections or promoting progression of active NTM infection isunknown. Untilmoreinformationisavailable, expertopinionis that patients with active NTM disease should receive TNF-blocking agents only if they are also receiving adequate therapyfor the NTM disease.LABORATORY PROCEDURESSincethepublicationofthelastATSstatementonNTM,theClinical andLaboratoryStandardsInstitute(CLSI), formerlyknown as the National Committee for Clinical Laboratory Stan-dards (NCCLS), has published an approved standard for NTMsusceptibility testing by mycobacteriology laboratories (42, 43).Theinstituteprovidesaglobal forumforthedevelopmentofstandards and guidelines. All proposed standards from the insti-tutearesubjectedtoanaccreditedconsensusprocessbeforebeing published as accepted standards. The institute suggeststhat to maintain efciency, effectiveness, and consistency in theinterpretationoftestresults, itisimportantthatotherhealthcareassociated organizations ascribe to the same standards andpracticesasapprovedbytheCLSI. Unlessnotedinthetext,recommendationsinthisdocument areconsistent withCLSIpublished standards.Collection, Digestion, Decontamination, andStaining of SpecimensSpecimens for mycobacterial identicationandsusceptibilitytestingmaybecollectedfromalmost anyareaof thebody.Collectionof all specimensshouldavoidpotential sourcesofcontamination, especially tap water, because environmental my-cobacteria are often present. Specimens should be submitted with-out xatives. Observing routine safety precautions by collectingsamplesinsterile, leak-proof, disposable, labeled, laboratory-approved containers is important. Transport media and preser-vatives are not usually recommended, although refrigeration ofsamplesat4Cispreferrediftransportation tothelaboratoryis delayed more than 1 hour. For diagnostic purposes, it may benecessary to collect multiple respiratory specimens on separatedays from outpatients. Specimens for mycobacterial analysis canbe shipped or mailed. Overnight shipping with refrigerants suchas coldpacks is optimal, althoughmycobacteriacanstill berecovered several days after collection even without these mea-sures. The longer the delay between collection and processing,however, the greater is the risk of bacterial overgrowth. Treat-mentwithcommonlyusedantibioticssuchasmacrolidesandquinolonesmightadverselyaffecttheyieldofNTMrecovery.Therefore, ifpossible, antibioticuseshouldbelimitedduringdiagnostic evaluation of NTM diseases.Respiratory SpecimensTo establish the diagnosis of NTM lung disease, the collectionof three early-morning specimens on different days is preferred.For patients unabletoproducesputum, sputumcanalsobeinduced. Induced sputum is an effective method for diagnosingTB; however, similar data establishing the effectiveness of spu-tum induction for diagnosing NTMlung disease are not available(44). In addition, the optimal methodology for sputum inductioninthissettinghasnotbeendetermined. Ifsputumcannotbeobtained, bronchoscopywithor without lungbiopsymaybenecessary. Becauseof clinical similaritiesbetweenNTMlungdisease and TB, appropriate precautions to prevent the nosoco-mialtransmissionofTBshouldbefollowedwhenperformingthese procedures. It is also important to performappropriate clean-ingproceduresforbronchoscopesthatincludetheavoidanceoftap water, which may contain environmental mycobacteria.Body Fluids, Abscesses, and TissuesAsepticcollectionof as muchbodyuidor abscess uidaspossiblebyneedleaspirationorsurgicalproceduresisrecom-mended. Swabs arenot recommendedfor samplecollectionbecausetheyoftenarenot aggressivelyapplied, resultinginlimited culture material, and are also subject to dessication, thusAmerican Thoracic Society Documents 373decreasing chances for recovery of NTM. If a swab is used, theswabshouldbesaturatedwiththesampleduidtoassureanadequate quantity of material for culture. When submitting tis-sue, the specimen should not be wrapped in gauze or diluted inliquidmaterial. Ifonlyaminuteamountiftissueisavailable,however, it may be immersed in a small amount of sterile salineto avoid excessive drying.BloodSeveral commercial mycobacterial bloodculturesystems forNTM are available (see online supplement). Coagulated bloodor blood collected in ethylenediaminetetraacetic acid (EDTA)is unacceptable. For rapidly growing mycobacteria (RGM), thesespecial mycobacterialsystemsarenotrequiredasmostRGMspecies grow well in routine blood culture systems.Specimen ProcessingTo minimize contamination or overgrowth of cultures with bacte-ria and fungi, digestion and decontamination procedures shouldbe performed on specimens collected from nonsterile body sites.Samples from contaminated sites contain other organisms thatmay grow more rapidly than NTM and interfere with the recov-ery of mycobacteria. Because NTM, especially RGM, are muchmore susceptible to decontamination than M. tuberculosis, theseprocedures should not be so severe as to eliminate the mycobac-teria potentially present. Tissue samples or uids from normallysterile sites do not require decontamination. Tissues should beground aseptically in sterile physiological saline or bovine albu-min and then directly inoculated onto the media.Themost widelyuseddigestiondecontaminationmethoduses N-acetyl-l-cysteinesodium hydroxide (NALC-NaOH). Thismethod is often used in conjunction with a 5% oxalic acid pro-cedure (double processing) for specimens frompatientswith CF or bronchiectasis whose sputa are known to be contami-nated with aerobic gram-negative rods, especially Pseudomonasaeruginosa (42, 45). Instructions for commonly used digestiondecontamination methods are described elsewhere (4648). Be-causeNTMmaybesensitivetooxalicaciddecontamination,withreducedyieldon culture, anotheroption istousea two-stepdecontaminationapproachreservingoxalicacidonlyforthose specimens overgrown by bacteria other than NTM(4749).Smear MicroscopyTherecommendedmethodforstainingclinical specimensforAFB, includingbothM. tuberculosisandNTM, istheuoro-chrome technique, although the Ziehl-Neelsen method orKinyoun stain are acceptable but less sensitive alternatives. TheGramstainisnotadequatefordetectionofmycobacteria. Inmany cases, the NTM, especially the RGM, may be more sensi-tive to the AFB decolorization procedure and may not stain atall with uorochrome stains. Therefore, if RGM are suspected,itmaybeprudenttouseaweakerdecolorizingprocess. Itisalso noteworthy that negative smearsdo notnecessarily meanthat NTM, especially RGM, are not present in a clinical sample.Semiquantitative analysis of smears can be useful for diagnos-ticpurposes. Fluorochromesmearsaregradedfrom1(19organisms per 10high-power elds) to4(90organismsper high-power eld) (47). The burden of organisms in clinicalmaterialisusuallyreectedbythenumberoforganismsseenonmicroscopic examination of stained smears. Environmentalcontamination, which usually involves small numbers of organ-isms, rarelyresultsinapositivesmearexamination. Previousstudieshaveindicatedthat specimenswithahighnumberofmycobacteriaisolatedbycultureareassociatedwithpositivesmears and, conversely, specimens with a low number of myco-bacteria isolated by culture are less likely to have positive smears(50).Culture TechniquesAll cultures for mycobacteria shouldinclude both solid and broth(liquid) media for the detection and enhancement of growth (43).However, brothmediaculturesalonemaynotbesatisfactorybecause of bacterial overgrowth. Cultures in broth media havea higher yield of mycobacteria and produce more rapid resultsthan those on solid media. The advantages of solid media overbroth media are that they allow the observation of colony mor-phology, growthrates, recognitionof mixed(morethanonemycobacterial species)infections, andquantitationof thein-fecting organism, and serve as a backup when liquid media cul-tures are contaminated.Brothmedia. One ofthe most widely usedbroth systems isthe nonradiometric mycobacteria growth indicator tube (MGIT)(BectonDickinson, Sparks, MD), whichcontains amodiedMiddlebrook 7H9 broth in conjunction with a uorescencequenchingbased oxygen sensor to detect mycobacterial growth.As the mycobacteria grow and deplete the oxygen present, theindicator uoresces when subjected to ultraviolet light. For de-tailed discussion of broth (liquid) media culture techniques, seethe online supplement.Solidmedia. Recommendedsolidmediainclude either egg-based media, such as Lo wenstein-Jensen agar or agar-based me-dia such as Middlebrook 7H10 and 7H11 media. The agar-basedmedia may also be used for susceptibility testing. Biphasic media,such as the Septi-Chek System (Becton Dickinson), provide en-hanced recovery of most NTM in one system, but these are notrapid detection systems.Semiquantitative (04) reporting of NTM colony counts onsolidmediais recommendedbytheCLSI. Asinglepositiverespiratory sample with a low colony count (e.g., broth culturepositiveonly) is less likelytobeclinicallysignicant thanasample with a high colony count (e.g., growth on both solid andbrothmedia). Thisapproachalsohelpsintheassessment ofresponse to therapy. After successful therapy of MAC lung dis-ease and at least 10 months of negative cultures during therapy,a single positive culture that is AFB smear negative and of lowculture positivity (10 colonies on solid media and/or positive inthe broth media only) generally represents either contamination(false-positiveculture)ortransientnewinfectionsandnotre-lapse of the original infecting strain (38, 51).M. haemophilum, M. genavense, M. avium subsp. paratuber-culosis (formerly M. paratuberculosis), and M. ulcerans are ex-amples of fastidious NTM that require special supplementationfor recovery on culture. M. haemophilum grows only on mediasupplementedwithiron-containingcompounds suchas ferricammonium citrate, hemin, or hemoglobin. Because M. haemo-philum has a predilection for skin and the bodys extremeties,all specimens fromskinlesions, joints, or bones shouldbecultured in a manner suitable for recovery of this species.M. genavense and M. avium subsp. paratuberculosis require my-cobactinJ, andM. ulceransmaybeoptimallyrecoveredwithegg yolk supplementation.Incubation of NTM CulturesThe optimal temperature for most cultures for NTM is between28and37C. Mostclinicallysignicantslowlygrowingmyco-bacteria grow well on primary isolation at 35 to 37C with theexception of the following: the newly described M. conspicuum,which requires temperatures from 22 to 30C for several weeksand only grows at 37C in liquid media, M. haemophilum, whichprefers temperatures from 28 to 30C, M. ulcerans, which growsslowlyat25to33C,andsomestrainsofM. chelonae,which374 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 2007requiretemperaturesbetween28 and33C(5). CulturesforRGM and M. marinum should be incubated at 28 to 30C. Allskin, joint uid, and bone specimens should be cultured at 28to 30C and at 35 to 37C. Optimal recovery of all species mayrequire duplicate sets of media at two incubation temperatures.Most NTM grow within 2 to 3 weeks on subculture. To detectM. ulceransor M.genavense, culturesshould be incubatedforatleast8to12weeks. Rapidlygrowingmycobacteriausuallygrowwithin7days of subculture. Earlier detectionof NTMcan be expected with liquid-based systems. When stated on thelaboratory report, the time in days to the detection of mycobacte-rial growth can be helpful to clinicians as an indication of isola-tion of a rapidly growing species.Recommendations:1. As much material as possible for NTM culture should beprovidedwith clear instructions to the laboratory toculturefor mycobacteria (C, III).2. All cultures for NTMshould include both a rapid detectionbroth(liquid)mediatechniqueandsolidmediacultures(C, III).3. Quantitation of the number of colonies on plated culturemediashouldbeperformedtoaidclinical diagnosis(C,III).4. Supplementedculturemediaandspecial culturecondi-tions (lower incubation temperatures) should be used formaterial cultured from skin lesions, joints,and bone (A,II).5. The time (in days) to detection of mycobacterial growthshould be stated on the laboratory report (C, III).NTM IdentificationBecause of differences in antimicrobial susceptibility that deter-mine treatment options, species-level identication of the NTMis becoming increasingly clinically important (43). Several factorsincrease the likelihood of clinical signicance of NTM isolates,including the recovery frommultiple specimens or sites, recoveryof the organism in large quantities (AFB smearpositive speci-mens), orrecoveryofanNTM isolate fromanormally sterilesitesuchas blood. For initial clinical mycobacterial isolates,however, it is sometimes difcult to determine the clinical sig-nicance of the isolate without species identication. Therefore,identication of most mycobacterial isolates to the species leveland not merely as groups, such as M. chelonae/abscessus groupis strongly recommended. If, after consultation between the clini-cian and the laboratorian and in the event that a specic labora-tory does not have the necessary technology for species identi-cation of an NTM isolate, the isolate could be sent to a referencelaboratory for further analysis.It is equallyimportant torecognizethat not all clinicallyobtained NTM isolates, especially from sputum, will need exten-siveidenticationefforts. For instance, apigmentedrapidlygrowingmycobacteriumrecoveredinlownumbersfromonlyoneofmultiplesputumspecimensfromapatientundergoingtherapyfor MAClungdiseasewouldnot needanextensiveeffort for identication of that isolate as it would not likely beclinicallysignicant. AwarenessofthecontextfromwhichanNTM isolate is obtained can be critically important in determin-ing the need for speciation of that isolate. Again, communicationbetween the clinician and laboratorian is critical for making thistype of determination.Phenotypictesting. Preliminarily, NTMcanbecategorizedby growth rate and pigmentation,which will help guide in theselectionof proper testingprocedures, includingappropriatemediaselectionandincubationtemperatures. Recent studieshave shown, however, that identication using only conventionalbiochemical analysis is both time consuming and increases turn-around time, leading to signicant delays in diagnosis (52).Detaileddescriptionsof methods, procedures, andqualitycontrol measures have been published (47, 48). Isolates of NTM,whichformcoloniesonsubculturein7daysorfewer, arere-ferredtoas rapidlygrowingmycobacteriaor RGM. Con-versely, those isolates of NTM that require more than 7 days toform mature colonies on subculture are termed slowly growingmycobacteria.Although manyoftheconventional andtradi-tional laboratory tests for mycobacterial evaluation are not rou-tinely used, the rate of growth is still useful for preliminary broadclassication of a nontuberculous mycobacterium.Traditionally, NTM have also been divided into three groupsbasedon the productionof pigment. These growth characteristicsarerarelydetailedinmodernmycobacterial laboratories, butthepresenceofpigmentationandsmoothcolonymorphologyquickly exclude the isolate as belonging to the M. tuberculosiscomplex that forms nonpigmented and rough colonies.Biochemical testingof NTMusesabatteryof tests, againbased on the growth rate of the NTM. The use of conventionaltesting alone does not allow identication of many of the newlydescribed NTM and thus newer methods, including HPLC andmolecular methods, must be used.Chemotaxonomic testing:HPLC. HPLC is a practical, rapid,and reliable method for identifying many slowly growing speciesof NTM. HPLC can also be used in the direct analysis of primaryculturesofmycobacteriagrowninBACTEC7H12Bmedium(Becton Dickinson) and the identication of MAC directly fromsamples with AFB smearpositive results (53). However, HPLChas limitations. Recognition of some newer species and specieswithintheM. simiaecomplexisdifcult. Ithasalsobeenre-ported that some species within the M. fortuitum group and theM. smegmatis grouparedifcult toseparate, anddistinctionbetween M. abscessus and M. chelonae may not be possible (54).HPLC analysis will be less useful in the future, as identicationof NTM species will be accomplished by molecular methods.Genotypic methods for identication of NTM. Commerciallyavailable molecular probes. AcridiumesterlabeledDNAprobesspecicforMAC(orseparatelyforM. aviumandM.intracellulare), M. kansasii, and M. gordonae have been approvedbytheU.S. FoodandDrugAdministration(FDA) andarecurrently used in many clinical laboratories (AccuProbe; Gen-Probe, Inc., San Diego, CA) for the rapid identication of NTM.This technique is based on the release of target 16S rRNA fromthe organism. Testing can be performed using isolates from solidorliquidculturemediaandidenticationofthesespeciescanbe achieved within 2 hours. Studies have shown 100% specicitywith a sensitivity between 85 and 100%. Only a few NTM specieshave available probes, representing a major limitation. In addi-tion, thereisthepotential forcross-reactionoftheprobeforM. tuberculosis with M. celatum (55).PRA. The current PRAmethod widely adoptedfor the identi-cation of NTM is based on the coupling of the PCR of a 441-bp sequence of the gene encoding the 65-kD heat shock protein(hsp65)followedbyrestrictionenzymedigestion. Thesizeofthe restriction fragments is generally species specic (5659). Inonestudy, 100%of 129nonpigmentedRGMwereidentiedusingPRA(60). However, sometaxamayrequireadditionalendonucleases for species identication (60).The PRA method is relatively rapid, does not require viableorganisms, and identies many NTM species that are not identi-able by phenotypic or chemotaxonomic techniques alone. How-ever, thissystemisnotcommerciallyavailable; therefore, theAmerican Thoracic Society Documents 375clinician may need to work closely with a public health or refer-ence laboratory to determine the best method for species identi-cation for a particular NTM isolate.DNAsequence analysis. The 16S rRNAis an approximately1,500 nucleotide sequence encoded by the 16S ribosomal DNA(rDNA), which is a highly conserved gene with regions commonto all organisms (conserved regions) and also areas where nucle-otide variations occur (variable regions). For purposes of myco-bacterial identication, sequence analysis focuses on two hyper-variable sequences known as regions A and B. The sequence ofregionAis usuallyadequatetoidentifymost NTMspecies,althoughsequencingofregionBmaybenecessary, especiallyin the identication of undescribed species or those species whichcannot be differentiated by sequence of the region A alone (5).Examples include M.kansasii/M. gastri, as well as M. ulceransand M. marinum, and M. shimoidei and M. triviale. Isolates ofM. chelonaeandM. abscessuscannotbedifferentiatedwithinthe regions A and B, although they do vary at other 16S rRNAgene sites (although only by a total of 4 bp) (60, 61).Problems with this method are that species of recent diver-gence may contain highly similar 16S rRNA gene sequences. Forexample, the difference between M. szulgai and M. malmoense istwo nucleotides, although it is well established that these are twodistinct species. In addition, no interstrain nucleotide sequencedifference value that unequivocally denes different species hasbeen established for mycobacteria (48).The automation of sequence analysis by the introduction ofcommercial systems like the MicroSeq 500 16S rDNA BacterialSequencingKit(PE AppliedBiosystems, FosterCity, CA),inwhich a 500-bp sequence of the NTM is analyzed and comparedwithacommerciallyprepareddatabase, hasmadesequencingmore efcient for use in the clinical laboratory. One of the majorlimitations of this system, however, is that the MicroSeq databasehas only one entry per species (generally the type strain) (61).This is particularly problematic when the unknown isolate doesnothaveanexactmatchinthedatabases. Currently, isolatesmay be reported as most closely related to a species dependingonthesequencedifferencebetweentheunknownisolateandthe database entry (62, 63).Recommendations:1. Clinicallysignicant NTMisolates shouldberoutinelyidentied to the species level. An important exception isMAC because the differentiation between M. avium andM. intracellulare is not yet clinically signicant. Althoughnotroutinelyrecommended, thisdifferentiationmaybeimportant epidemiologically and, in the future, therapeuti-cally (C, III).2. The RGM(especially M. chelonae, M. abscessus, andM. fortuitum)shouldbeidentiedtospecieslevelusingarecognizedacceptablemethodology, suchas PRAorbiochemical testing, not HPLC alone (A, II).3. Susceptibility of RGM for eight agents, including amikacin,cefoxitin, clarithromycin, ciprooxacin, doxycycline, linez-olid, sulfamethoxazole, and tobramycin, can also be usedtofacilitateidenticationof M. abscessus, M. chelonae,and M. fortuitum (C, III).4. Communicationbetweentheclinicianandlaboratorianis essential for determining the importance and extent ofthe identication analysis for a clinical NTMisolate(C, III).Antimicrobial Susceptibility Testing for NTMContext:There is ongoing debate about the role of in vitro susceptibilitytesting for managing patients withNTMdisease. The controversyprimarily stems fromthe observation that, unlike M. tuberculosis,MACresponsetoanti-TBdrugssuchasrifampinandetham-butol maynot bereliablypredictedonthebasis of currentinvitrosusceptibilitytestmethods. NTMsuchasM. kansasii,M. marinum, or M. fortuitum are susceptible in vitro to multipleantimicrobial drugs andthe clinical response totherapeuticagents appears to closely parallel the in vitro susceptibility pat-tern, although this observation has not been evaluated by ran-domized controlled trials. In contrast, MAC has limited in vitrosusceptibility, and clinical response has been shown to correlateonly with the macrolides. Last, organisms such as M. abscessusandM.simiaehave limitedinvitrosusceptibility,with limitedevidence for a correlation between in vitro susceptibility to anyagent and clinical response in the treatment of pulmonary diseasecausedbytheseagents. Interestingly, forskinandsoft tissueinfectionscausedbyM. abscessus, theredoesappeartobeacorrelation between in vitro susceptibility and clinical response,although this observation has not been prospectively evaluated.Susceptibility breakpoints have been dened in the laboratoryto distinguish populations of mycobacteria that are labeled sus-ceptible and resistant. For many NTM, however, these labora-tory cutoffs have not been conrmed to be clinically meaningful,sothattherearefew datatovalidate susceptibilitytesting formany NTM species as a guide for choosing antibiotics. Until therelationshipbetweeninvitrosusceptibilityofmanyNTMandtheir clinical response to antimicrobial drugs is better understoodand claried, the clinician should use in vitro susceptibility datawithanappreciationofitslimitationsandwiththeawarenessthat, unlike TB, some NTM disease may not be eradicated in agiven patient with therapy based on in vitro susceptibility results.TheuseofinvitrosusceptibilitytestingforNTMisoutlinedbelow and consistent with CLSI recommendations.Slowlygrowingmycobacteria. FortheslowlygrowingNTM,no single susceptibility method is recommended for all species.For isolates of MAC, the CLSI has recommended a broth-basedmethod with both microdilution or macrodilution methods beingacceptable (43). Until further multicenter studies have been per-formed with the other slowly growing mycobacteria, broth andsolidmethodsofsusceptibilitytestingmaybeperformedwiththe caveat that each laboratory must validate each method foreach species tested, and quality control and prociency testingrequirements should be enforced.MAC. Initial isolatesfrompatientswithpreviouslyuntreatedMAClung disease should be tested to clarithromycin to establishbaseline values. The isolate may also be saved for future testing,especiallytodetermineiffutureisolatesrepresentrelapseornew infections by comparative DNA ngerprinting of baselineand subsequent isolates (51). Other isolates to be tested includethe following:1. Isolates from patients who previously received macrolidetherapy to determine whether or not the isolates are stillmacrolide susceptible.2. IsolatesfrompatientswithMACpulmonarydiseaseonmacrolide-containing regimens who relapse or fail after 6months of macrolide-containing therapy.3. Isolates from patients with AIDS who develop bacteremiaon macrolide prophylaxis.376 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 20074. Positive blood cultures after3 months of treatment withmacrolide-containing regimens for patients with dissemin-ated MAC.Untreated MAC isolates usually have minimum inhibitory con-centrations (MICs) of 4 g/ml or less to clarithromycin and areconsideredsusceptible. Incontrast, relapsestrainsaftertreat-ment inevitably have a clarithromycinMICof 32 g/ml or greaterandnolongerrespondtotreatmentwithmacrolides. IsolatesofMAChaveonlyasinglecopyoftheribosome, andhence,macrolide monotherapy carries a signicant risk of the develop-ment of mutational resistance. Not unexpectedly, all high-levelclarithromycin-resistant isolates have mutations in the adenineatposition2058or2059ofthe23SrRNAgene, whichisthepresumed macrolide binding site on the ribosomal unit (64, 65).Expertconsultationshouldbesoughtformanagementofpa-tients with macrolide-resistant MAC isolates.Strains that appear intermediate in susceptibility to clarithro-mycinrarelyoccurandshould beconrmed byanother testingevent. Patients with these intermediate MICs should be followedclosely for possible emergence of macrolide resistance. Macrolidesshould be included in treatment regimens for these patients unlessthe isolate is found onsubsequent testing to be macrolide resistant.Because no correlation between in vitro susceptibility resultsfor MAC and clinical response for agents other than macrolideshas been established, the 2003 CLSI document states that clar-ithromycin istheonlydrug forwhichsusceptibility testing forMAC isolates is recommended (43).The role of extended in vitro susceptibility testing for macro-lide-resistant MAC isolates is unproven. However, some expertssuggestthatitmaybereasonabletotestotherantimicrobialssuch as the 8-methoxy uoroquinolone moxioxacin and linezo-lidforpatientswhofail initial macrolide-basedtherapy(43).Although there is a paucity of published data, some experts feelthat moxioxacin has in vitro activity against MAC at clinicallyachievable serum levels. It was recently reported that out of 189isolates of MAC, only 13% of the isolates studied had MICs of8g/ml or less for linezolid; however, there have been reportsof successful multidrug therapy with linezolid in the treatment ofMAC (66, 67). Testing of anti-TB medications does not provideuseful clinical information.M. kansasii. The 2003 CLSI document states that only suscep-tibilitytorifampinshouldberoutinelyperformedforisolatesof M. kansasii (43). MICs of all agents in untreated (wild) strainsfall in a narrow range and treatment failure is usually associatedwith resistance to rifampin. Resistance to isoniazid and ethambu-tol acquired on therapy may also occur, but resistance to theseagents is usuallyassociatedwithresistancetorifampin(68).Isolates that are susceptible to rifampin are also susceptible torifabutin, which may be substituted for rifampin in HIV-infectedpatients being treated with highly active antiretroviral therapy(HAART), including some protease inhibitors and nonnucleo-sidereversetranscriptaseinhibitors(seeDisseminatedMACDisease). If the isolate proves to be rifampin resistant, suscepti-bilitytosecondaryagents, includingamikacin, ciprooxacin,clarithromycin, ethambutol, rifabutin, streptomycin, sulfon-amides, and isoniazid, should be tested. Susceptibility to the new8-methoxy uoroquinolone, moxioxacin, should be performedseparately becauseciprooxacinisthe classrepresentativeforciprooxacin, ooxacin, and levooxacin only. In addition, thestandard critical concentrations of 0.2 and 1 g/ml for isoniazidused for testing strains of M. tuberculosis should not be testedasMICs foruntreatedstrains between0.5and5g/ml.Thus,the 0.2-g/ml standard concentration appears resistant and the1.0-g/ml standardconcentrationmay yieldvariable results,even with multiple cultures of the same strain (43).M. marinum. Routine susceptibility testing of this species isnot recommended (43). There is no documentation of signicantrisk of mutational resistance to the antimycobacterial agents andthereisnoappreciablevariabilityinsusceptibilitypatternstoclinicallyuseful agents(69). DiseaseinvolvingM. marinumistypically localized and the number of organisms present is low;95% of tissue biopsy cultures are AFB smear negative (2). Iso-lates ofM.marinumaresusceptibletoclarithromycin,as wellas the sulfonamides, the tetracyclines, rifampin, and ethambutol.Ciprooxacinis not recommendedbecausesomestrains areresistant and monotherapy carries the risk of mutational resis-tance (70). However, some experts report anecdotally that thenewer 8-methoxy uoroquinolone, moxioxacin, is more activein vitro and could be considered for multidrug therapy. Suscepti-bility testing should be considered for patients who remain cul-ture positive after more than 3 months of therapy.Miscellaneous slowly growing NTM (M. simiae complex,M. terrae/nonchromogenicumcomplex, M. malmoense, M.xenopi). Becausetoofewisolates of eachspecies havebeenstudied, no specic susceptibility method can be recommendedat this time for the less commonly isolatedNTMincluding severalnewly described species (5, 43). Until further data are available,testing should be performed as for rifampin-resistant M. kansasii(i.e., rifampin and secondary agents should be tested) (43).RGM. TheCLSIhasrecommendedthebrothmicrodilutionMIC determination for susceptibility testing of RGM. Agar tests,includingtheE-test(epsilometertest), cannotberecommendeddue to inconsistency of results (71). The broth microdilution tech-nique is described in the 2003 NCCLS M24A-approved document(40). MICs for imipenemare problematic with isolates of M. chelo-nae, M. abscessus, and M. immunogenum because of the lack ofreproducibility (43, 72). For the majority of M. abscessus and M.chelonae isolates, imipenem is the preferred carbapenemover mer-openemandertapenem(73). Imipenemmay still be useful clinicallyin treatment regimens for these organisms. In contrast, MICs forimipenem with isolates of the M. fortuitum group, M. smegmatisgroup, and M. mucogenicum are reproducible.Trueaminoglycosideresistancewith M.abscessusis unusualbut does occur, especially in patients on long-term treatment withaminoglycosides, such as patients with CF or chronic otitis media.In vitro susceptibility studies suggest that tobramycin is the mostactiveaminoglycosidefor M. chelonae; therefore, it is recom-mended to report tobramycin MICs only for this species (43).In vitro testing of clarithromycin may present interpretationproblemswithRGM. Thus, theCLSIhasrecommendedthatisolates of the M. fortuitum group with indeterminate or unclearsusceptibility endpoints with clarithromycin should be reportedasresistantuntilfurtherdataareavailablewiththeseisolates(43). Recent studies have shown that all isolates of M. fortuitumcontainaninducibleerythromycinmethylasegeneerm(39),whichconfersmacrolideresistance(74). Thepresenceofthisgene with variable expression is likely responsible for this phe-nomenon. Similar erm genes have been reported in other RGMspecies that are macrolide resistant (e.g., M. smegmatis) but notin M. abscessus despite its general poor response to macrolidetherapy (75).Fastidious species of NTM. As for the previous group of lesscommonly recovered NTM, no current standardized method canberecommendedat thistimeduetolackof experienceandavailable data concerning methods and results.M. haemophilum. Isolatesaregenerallysusceptibletotherst-lineanti-TBagents (except ethambutol), clarithromycin,and the sulfonamides (76).M. genavense, M. avium subsp. paratuberculosis,M.ulcerans. Susceptibilitytestingofthesespeciesisdifcultsince they do not grow in standard susceptibility media withoutAmerican Thoracic Society Documents 377supplementation and extended incubation; therefore, standard-ized guidelines for in vitro susceptibility procedures are not avail-able for testing these species (7782).Recommendations:1. Clarithromycin susceptibility testing is recommended fornew, previously untreated MAC isolates. Clarithromycinisrecommendedastheclassagentfortestingof thenewer macrolides because clarithromycin and azithro-mycin share cross-resistance and susceptibility. No otherdrugs are recommended for susceptibility testing of new,previously untreated MACisolates. There is no recognizedvaluefortestingofrst-lineantituberculous agentswithMAC using current methodology (A, II)2. Clarithromycin susceptibility testing should be performedfor MAC isolates from patients who fail macrolide treat-ment or prophylaxis regimens (A, II).3. Previously untreated M. kansasii strains should be testedin vitro only to rifampin. Isolates of M. kansasii that showsusceptibility torifampinwill alsobesusceptibleto rifa-butin (A, II).4. M. kansasii isolates resistant to rifampin, should be testedagainstapanel ofsecondaryagents, includingrifabutin,ethambutol, isoniazid, clarithromycin, uoroquinolones,amikacin, and sulfonamides (A, II).5. M. marinum isolates do not require susceptibility testingunless the patient fails treatment after several months (A,II).6. Therearenocurrent recommendationsforonespecicmethod of in vitro susceptibility testing for fastidiousNTM species and some less commonly isolated NTM spe-cies (C, III).7. Validation and quality control should be in place for sus-ceptibility testing of antimicrobial agents with all speciesof NTM (C, III).Molecular Typing Methods of NTMMolecular typing methods have become valuable epidemiologictoolsintheinvestigationofoutbreaks,pseudo-outbreaks,andepidemics involving NTM. If a laboratory that performs molecu-lar typing methods of NTM isolates cannot be readily identied,theclinician orinvestigatorshouldcontactthe statesDepart-ment of Health or the CDC for advice about laboratories thatcan assist with this analysis.Pulsed-eld gel electrophoresis. Pulsed-eld gel electrophore-sis (PFGE) is one of the most widely used and valuable methodsof molecular typing of NTM. This technique involves embeddingthe isolates in agarose gels, lysing the DNA, and digesting chro-mosomal DNA with specic restriction endonucleases (7981).This is a time-consuming procedure because the organisms mustbe actively grown such that a sufcient biomass is available toyield accurate results. Furthermore, more than one-half of theDNA from strains of M. abscessus will spontaneously lyse or bedigestedduringtheprocedure, althoughthiscanbecorrectedwiththeadditionof thioureatostabilizetherunningbuffer(7981).Despite its limitations, however, PFGE is currently the mostcommon typing method for strain differentiation of RGM andother NTM. Unrelated strains of most species of RGMare highlyheterogeneousandrestrictionfragment lengthpolymorphism(RFLP)patternsforthesamestrainareidentical orindistin-guishable (8386).Other typingmethods. Other methods havebeenusedforstrain comparison, including randomamplied polymorphicDNAPCR, multilocusenzymeelectrophoresisusingmultiplehousekeeping cellular enzymes, and hybridization with multicopyinsertional elements in M. avium but not M. intracellulare.CLINICAL PRESENTATIONS ANDDIAGNOSTIC CRITERIAPulmonary DiseaseEpidemiology. Chronicpulmonary disease is the most commonclinical manifestation of NTM (19,87, 88). MAC, M. kansasii,andM. abscessus,in that order, werethe most frequent NTMpulmonarypathogensintheUnitedStatesbetween1981and1983(19). Inthat study, therewasapredominanceof maleswith NTM pulmonary disease in general and in disease causedby all species except M. chelonae (probably M. abscessus) andM. simiae (19). The mean age of patients with NTM pulmonarydisease was 57 years (19). In the CDCreport fromthe mid-1990s,MAC, M. kansasii, andM. fortuitumwerethemostfrequentpulmonarypathogensintheUnitedStatesbetween1993and1996 (22). Males predominated in disease caused by all speciesexcept M. abscessus and the majority of isolates were from pa-tients 50 years of age or older. In light of more recent informationthat reects a postmenopausal female patient predominance forMAC lung disease, it is likely that these epidemiologic estimatesare not currently valid (36, 37).Symptoms and signs. The symptoms of NTM pulmonary dis-ease are variable and nonspecic. However, virtually all patientshave chronic or recurring cough.Other symptoms variablyin-cludesputumproduction,fatigue, malaise,dyspnea,fever,he-moptysis, chest pain, and weight loss. Constitutional symptomsare progressively more prevalent with advancing NTM lung dis-ease. Evaluationisoftencomplicatedbysymptomscausedbycoexisting lung diseases, such as bronchiectasis, chronic obstruc-tive airway disease associated with smoking, CF, and pneumoco-niosis.Physical ndings are nonspecic and reect underlying pul-monary pathology, such as bronchiectasis and chronic obstruc-tivelungdisease. Onchestauscultation, ndingsmayincluderhonchi, crackles, wheezes, and squeaks. Patients with nodular/bronchiectatic MAC disease tend to be postmenopausal women,manyof whomalsohaveacharacteristicmorphotypewithathin body habitus and may also have scoliosis, pectus excavatum,and mitral valve prolapse (27).Radiographic studies. Radiographic features of NTM lung dis-ease depend on whether the lung disease is primarily brocavi-tary (similar to TB) or characterized by nodules and bronchiecta-sis (nodular/bronchiectatic disease) (see the online supplement).ComparedwiththeradiographicndingsinTB, patientswithNTMdisease and predominantly brocavitary radiographicchanges tendtohavethefollowingcharacteristics: (1) thin-walled cavities with less surrounding parenchymal opacity, (2)lessbronchogenicbutmore contiguousspreadofdisease,and(3)toproducemoremarkedinvolvementof pleuraovertheinvolved areas of the lungs (2, 88, 89). None of these differences,however, is sufciently specicto excludethe diagnosis ofTBon the basis of the radiographic appearance. NTM may producedense airspace disease or a solitary pulmonary nodule withoutcavitation. Basal pleural disease is not often found, and pleuraleffusion is rare.For patients with predominantly noncavitary disease, the ab-normalities on chest radiograph are primarily found in the mid-andlower lungeld. Studies withHRCTof thechest haveshownthat upto90%of patientswithmid- andlowerlungeld noncavitary disease with MAC have associated multifocal378 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 2007bronchiectasis, with many patients having clusters of small(5 mm) nodules in associated areas of the lung (9094). Thesendings correspond histopathologically to bronchiectasis, bron-chiolarandperibronchiolarinammation, andgranulomafor-mation (94). Cavitation also frequently accompanies theseabnormalities. Other NTMspecies, including M. abscessus,M. chelonae, M. simiae, andM. kansasii (andprobablyotherspecies),canalsobe associatedwith thisradiographic appear-ance (32, 95).Aplainchest radiographmaybeadequateforevaluatingpatients with brocavitary disease. However, HRCT of the chestis now routinely indicated to demonstrate the characteristic ab-normalities of nodular/bronchiectatic NTM lung disease.Mycobacterial cultures. Presumptive diagnosis based on clini-cal andradiographicfeaturesisnotadequateforinitiationoftherapy. Isolation of NTM in culture is essential for the diagnosisof NTM lung disease. It should be stressed that NTM are foundwidely in nature; therefore, contamination of respiratory speci-mensoccurs. Asinglepositive sputumculture, especiallywitha small number of organisms, is generally regarded as indetermi-nant for diagnosis of NTM lung disease. NTM species that aregenerally not pathogenic and usually isolated due to contamina-tion when recovered from respiratory specimens include M. gor-donae, M. terrae complex, M. mucogenicum, andM. scrofu-laceum(Table2). Otherspeciesknowntobepresent intapwaterthatmayreectcontaminationwhenrecoveredfromasingle sample include M. simiae and M. lentiavum.Patientsshouldhaveat least threesputumspecimenscol-lected on separate days and analyzed for AFB to optimize posi-tivepredictive valueofsputumanalysis. Inastudyevaluatingthe association of MAC isolated from sputum and new cavitaryor inltrative lesions on chest radiograph (96), 114 patients hada single isolation of MAC (from three specimens) and only 2 ofthese patients, both with specimens that were AFB smear posi-tive, subsequently developed new chest radiographic abnormali-tites. In addition, 26 of 29 (90%) patients who had MAC isolatedfromtwospecimensand39of 40(98%)patientswithMACisolated from three specimens had progressive radiographic ab-normalities. All 116 patients who had four or more MACisolateshadprogressiveradiographicabnormalities. Furthermore, 181of 185(98%) patients whohadtwoor moreMACisolates,generally on the initialthree sputum specimens collected, alsohad progressive radiographic abnormalities.Clinical studies haveestablishedthevalidityof bronchialwashingsasaculturesourceforM. tuberculosis(44).Limiteddata suggest that bronchial lavage may also be useful for diagnos-ing NTM (MAC) lung disease (97).There is expert consensusthat bronchial washings are more sensitive than routine expecto-rated sputum testing and less likely to be affected by environ-mental contamination if the bronchoscopic specimens areprotectedfromtapwater (seeHealthCareandHygiene-associated Disease and Disease Prevention). The routine useof bronchoscopyfordiagnosisandfollow-upofpatientswithNTM lung disease is not established.The recovery of NTM from sputum can sometimes obfuscateor delay the diagnosis of other important lung diseases (98, 99).Inpatientswithnondiagnosticmicrobiologicandradiographicstudies (i.e., patients who do not clearly meet diagnostic criteria),orif thereis concernabout thepresenceof anotherdiseaseproducing radiographic abnormalities, a lung biopsy (broncho-scopicorsurgical)mayberequiredfordiagnosis. If atissuesample froma transbronchial, percutaneous, or open-lung biopsyyields anNTMorganismandshows histopathologicchangestypical of mycobacterial disease (i.e., granulomatous inammationwith or without the presence of AFB), this by itself is sufcientto establish the diagnosis of NTM lung disease. If the lung biopsyis negativeonculture(whichmayoccur whentransbronchialbiopsiesareperformedbecauseof thesmall sizeof thetissuesample) but demonstrates mycobacterial histopathology features(without a history of other granulomatous or mycobacterial dis-ease), NTM lung disease is considered to be present when oneor more sputum specimens or bronchial washes are culture posi-tive for NTM.Othertests. Therehasbeenagreatdeal ofinterestintheavailability of species-specic skin test antigens. Unfortunately,many antigenic epitopes are shared by different mycobacterialspecies and extensive cross-reactions are observed with differentmycobacterial skintest antigens. Dual skintestingwithPPDtuberculin and M. avium sensitin can help discriminate betweenculture-positivelungdiseaseduetoMACandthatduetoM.tuberculosis (100). However, M. avium sensitin is not being de-veloped for commercial use as an intradermal skin test.Recommendations:1. The minimum evaluation of a patient suspected of NTMlung disease should include (1) chest radiograph or, in theabsence of cavitation, chest HRCT scan; (2) three or moresputum specimens for AFB analysis; and (3) exclusion ofother disorders such as TB and lung malignancy. In mostpatients, a diagnosiscan bemade without bronchoscopyor lung biopsy (A, II).2. Disease caused by M. tuberculosis is often in the differen-tial diagnosis for patients with NTM lung disease. Empirictherapy for TB, especially with positive AFB smears andresults of nucleic acid amplication testing, may be neces-sary pending conrmation of the diagnosis of NTM lungdisease (C, III).Diagnosticcriteria. Overlyrigorouscriteriamight delayorprevent the diagnosis, with the subsequent risk for progressivedisease. Conversely, criteriathat aretoolenient couldresultinunnecessaryexposureof patients topotentiallytoxic andexpensive therapy. Because NTMlung disease is generally slowlyprogressive(relativetoTB),thereisusuallysufcienttimetocollectadequateclinicalmaterial,specically multiplerespira-tory specimens, necessary for making a diagnosis. For patientsin whom the diagnosis is unclear, expert consultation should besought.Given the large number of identied NTM species, the widespectrum of NTM virulence, and the variable host susceptibilityforNTM, itisunlikelythatasinglesetofdiagnosticcriteriawouldbe useful oraccurate for all NTMspecies in allclinicalcircumstances. Alimitation of all diagnostic criteria developed sofar is that, by necessity, they were developed based on experiencewithcommonandwell-described respiratorypathogenssuchasMAC, M. kansasii, and M. abscessus. It is assumed, but not proven,thattheconceptsoutlinedintheseguidelinesarepertinentforother less common NTMrespiratory pathogens. Suggested criteriafor diagnosing NTM lung disease are listed in Table 3.The previous ATSguidelines includedrecommendations basedon quantitation of smears and cultures in the diagnostic criteria.Many laboratories, however, do not report quantitative smear andcultureresults,especiallythoseusingonlyliquid(broth)culturemedia. Therefore, the diagnosis of NTM lung disease must some-timesbemadeonthebasisof smearandculturepositivityornegativity without quantitation. As noted in Laboratory Proce-dures, both liquid and solid media cultures are recommended, asis quantitation of mycobacterial growth on solid media cultures.It hasbeenpreviouslysuggestedthat therespiratorytractcan be infected with NTMwithout disease, particularly inAmerican Thoracic Society Documents 379TABLE 3. CLINICALANDMICROBIOLOGICCRITERIAFORDIAGNOSINGNONTUBERCULOUSMYCOBACTERIALLUNGDISEASE*Clinical (both required)1. Pulmonary symptoms, nodular or cavitary opacities on chest radiograph, or a high-resolution computed tomography scan that shows multifocal bronchiectasiswith multiple small nodules (A, I)*and2. Appropriate exclusion of other diagnoses (A, I)Microbiologic1. Positivecultureresultsfromatleasttwoseparateexpectoratedsputumsamples(A,II).Iftheresultsfrom(1)arenondiagnostic,considerrepeatsputumAFBsmears and cultures (C, III).or2. Positive culture result from at least one bronchial wash or lavage (C, III)or3. Transbronchial or other lung biopsy with mycobacterial histopathologic features (granulomatous inflammation or AFB) and positive culture for NTM or biopsyshowing mycobacterial histopathologic features (granulomatous inflammation or AFB) and one or more sputum or bronchial washings that are culture positivefor NTM (A, II)4. Expert consultation should be obtained when NTM are recovered that are either infrequently encountered or that usually represent environmental contamination(C, III)5. Patients who are suspected of having NTM lung disease but do not meet the diagnostic criteria should be followed until the diagnosis is firmly established orexcluded (C, III)6. Making the diagnosis of NTMlung disease does not, per se, necessitate theinstitution of therapy, which is a decision basedon potential risks and benefits oftherapy for individual patients (C, III)* For evidence quality, see Table 1.patients with chronic respiratory disease (18, 19, 84). This condi-tion has been referred to as colonization, and was describedmostoftenwithMAC.Theconceptofairwaycolonization byNTMhasneverbeentestedrigorously. Thereisnot enoughknownaboutthepathophysiologyofNTMlungdiseasetobesure that colonization is not, in fact, indolent or slowly progres-sive infection. No pathologic studies have been done to demon-stratetheabsenceoftissueinvasion, andmorerecentstudieswith HRCT have shown that these patients often have a combi-nationof multifocal bronchiectasis andnodular parenchymaldisease nowbelieved to be due to mycobacterial disease (9094).Colonizationwithout infection(i.e., notissueinvasion) isanunproven condition for NTM. Given the generally slow progres-sionofNTMlungdisease, itisincumbentonthecliniciantocollectenoughrespiratoryspecimensforAFBanalysisandtofollowpatients for an adequate period of time for conrmationorrefutation of the diagnosis of NTM lung disease. If the diagnosisremains in question, the patient should remainunder observationand expert consultation sought.Because NTM can be isolated due to environmental contami-nation, including contamination of clinical specimens, in generalmore than one culture-positive specimen for NTM is necessaryfor diagnostic purposes. The exception, outlined in the diagnosticguidelines above, is the patient with classic symptoms and radio-graphicndingsfornodular/bronchiectaticNTMlungdiseasewhoisunabletoproducesputumforAFBanalysis. Forthisspecic type of patient, the isolation of NTM, especially MAC,fromonebronchoscopicspecimenisconsideredadequateforthe diagnosis of NTM lung disease. For most NTM other thanMAC, expert consultation would be required to determine thesignicance of NTM isolated from a single bronchoscopic speci-men. The signicance of a single sputum specimen culture posi-tive for a nontuberculous mycobacterium is more uncertain. Ina study of 118 men in a gold-mining workforce in South AfricafromwhomNTMwereisolated, only27%mettheATScasedenitionsforpulmonaryNTMdisease(2, 101). Thispatientpopulation had a high incidence of TB and HIV infection; there-fore, most patients (70%) were started on empiric anti-TB ther-apybeforespeciesidenticationof themycobacterial isolate.M. kansasii, which is susceptible to anti-TB drugs, was the mostcommonNTMspeciesisolated. Formanypatients, therefore,follow-up (conrmatory) respiratory AFB specimens had nega-tive results. The authors concluded that (1997) ATS case deni-tions were difcult to apply in their high-risk study population.For selectedpatients, therefore, treatment decisions canbeguided by the pathogenic potential of NTM isolates, especiallyavirulent NTMspeciessuchasM. kansasii, evenif multiplespecimens are not positive or available. Expert consultation maybe helpful for making this decision. In most circumstances andfor most NTM respiratory isolates, however, one positive cultureis not adequate for making a diagnosis of NTM lung disease.Conversely, there may also be circumstances where patientsmeetdiagnosticcriteriaforNTMlungdiseasebut donot, infact, haveprogressivediseaseorsufcientlyseverediseasetowarranttherapy.Suchacircumstancecouldarisewiththere-peatedisolationoflow-virulenceNTM, suchasM. fortuitum,orNTMusuallyassociatedwithspecimencontamination(seeTable 2). These patients require careful clinical evaluation andcollection of multiple specimens for AFB analysis over time and,sometimes, more invasive diagnostic procedures. A strength ofthe diagnostic guidelines is the emphasis on longitudinal patientfollow-up, anopportunityprovidedbythegenerallyindolentnature of NTMdisease. Most diagnostic uncertainty can be over-come with this approach, including determining the signicanceof therecoveryof relativelylessvirulent NTMspecies(e.g.,M. fortuitum), and NTM species that are usually contaminants(M. gordonae and M. terrae complex). This determination mayalso require expert consultation.The interpretation of NTM in the sputum of HIV-infected pa-tients presents a particular problem. In general, for patients withabnormal chest radiographs, the diagnostic criteria recommendedfor immunocompetent hosts are still applicable, with an emphasison the exclusion of other possible pulmonary pathogens. In theabsence of radiographic evidence of pulmonary disease, respira-tory isolates of NTM in HIV-seropositive persons may be due todisseminated NTMdisease or can be a harbinger of disseminateddisease or represent transient infection (102). In addition, someNTM species that are generally considered nonpathogenic havebeenassociatedwithpulmonarydiseaseintheHIV-infectedhost. Given these considerations, the diagnosis of lung diseasecaused by NTMis usually not difcult if a combination of clinical,radiographic, and bacteriologic criteria is used.Last, thereareclinical problemsnotdirectlyaddressedbythese diagnostic guidelines. For instance, the signicance of an380 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 175 2007NTM isolated from a patient during therapy for pulmonary TBis uncertain. The signicance of two NTM species isolated simul-taneously from a patient is also unknown. The combination ofMAC and M. abscessus is especially well recognized. Unfortu-nately, there is not sufcient information to answer these issuesbroadlysothat patients inthesecircumstances must beap-proached on an individual basis. Both these events are likely tooccurwithincreasedfrequencybecauseofimprovedrecoveryof NTM by mycobacteriology laboratories. Patients who presentwith these clinical scenarios must be evaluated carefully, on anindividual basis, and may require expert consultation.Cystic FibrosisEpidemiology. Before 1990, NTM were uncommonly associatedwith CF (103). Since that time, multiple centers across the UnitedStatesandinEuropehavereportedaprevalenceofNTMinrespiratory specimens ranging from 4 to 20% in the CF popula-tionscreened. Inarecent largecross-sectional assessment ofNTM prevalence in the United States, three respiratory speci-mensweresystematicallycollectedoverthecourseof ayearfrom approximately 1,000 subjects aged 10 years and older from21 geographically dispersed CF centers (12). NTM were recov-ered from at least one of these specimens in 13% of the subjects.Most of these had a single positive culture (70%), but 16% hadtwoand13%hadthreepositivecultures. Smearresultswerepositive in 26%of culture-positive specimens. The most commonspeciesrecoveredwereMAC(76%)andM.abscessus(18%),withafewsubjects(4%)having bothspecies.TheprevalenceofNTMseemstobehighlycorrelatedwithage, approaching40%inpatients older than40years (12), althoughpediatricpatients have a higher proportion of M. abscessus and are morelikely to meet the ATS microbiologic criteria for disease (104,105).Pathophysiology. The reasons for the high prevalence of NTMinpatients withCFarenot clear. Theunderlyingstructuralairway disease and altered mucociliary clearance may be predis-posing factors. A similar prevalence of NTM has been describedin primary ciliary dyskinesia, also characterized by bronchiectasisandalteredmucociliaryclearance(106). Anincreasedpreva-lence of mutations in the CF gene (CF transmembrane conduc-tance regulator [CFTR]) has been noted in elderly persons withNTMand nodular bronchiectatic lung disease, suggesting a possi-bleroleof CFTRinpredisposingtoNTMinfection(3335).ThetransmissibilityofotherpathogensbetweenpatientswithCF, such as Burkholderia cepacia, and the frequent aggregationof patients onhospital wards for prolongedperiods of timeraisequestionsaboutperson-to-persontransferornosocomialacquisitionfrominstitutional watersupplies. Twosingle-centerand a large multicenter study using molecular epidemiologic tech-niqueshavefailedtoshowanyevidenceof person-to-persontransfer (12, 104, 107). Similarities in frequency of clinic visits,dayshospitalized, andaerosol inhalational deviceuseamongpatients with CF with and without positive cultures suggest nogreater exposure to medical water reservoirs among NTM-posi-tivepatientswithCF(12, 104). Althoughnotlikelycommonsources of acquisition, institutional water reservoirs remain po-tential sources of concernaswas notedinarecent studyofan M. simiae outbreak, which included a patient with CF withmultiple smear-positive isolates (108).Diagnosis. Ingeneral, thediagnosticcriteriaforNTMlungdisease listed earlier in this statement are applicable for patientswith CF as well. However,diagnosing disease caused by theseorganismscanbequitedifcultduetooverlappingsymptomsand radiographic changes attributable to the underlying CF. Itcan be difcult to exclude other causes given the frequent pres-ence of other organisms such as P. aeruginosa and Staphylococ-cusaureus, whichareassociatedwithsimilarclinical disease.Although there is overlap, the ndings on HRCT scans associ-ated with nodular bronchiectatic disease may be distinguishablefrombackgroundndingsofCF(29, 92, 109, 110)(seeonlinesupplement).Treatment. Although the majority of patients with CF fromwhomNTMarerecoveredwill likelynotmeetmicrobiologiccriteria for disease at the time of an initial positive culture, closesurveillanceof thesepatients is warranted. Therehavebeennumerous reports of clinical deterioration and death temporallyassociated with persistent recovery of these organisms, particu-larly heavy growth of M. abscessus from lower respiratory speci-mens (111120). It is important that nonmycobacterial pathogensbe maximally treated before initiating specic antimycobacterialtreatment, given the overlapping spectrum of antimycobacterialdrugs for common CF pathogens, to facilitate assessment of theclinical response to antimycobacterial treatment. Assessment ofthe effect ofNTM treatment on clinicalparameters like FEV1may be further confounded by the possible immunomodulatoryeffect of the macrolides in CF (121). A potential consequence ofthebroadadoptionofthelow-doseandlong-term azithromycintherapy in patients with CF is the evolution of macrolide-resistantNTM. A careful evaluation for possible pulmonary NTM infec-tion, including multiple sputum cultures for NTM, should pre-cede any initiation of macrolide monotherapy, and cultures forNTM should be obtained periodically thereafter. Obtaining anHRCTscan ofthe chest and serial measures of lung function,weight, and other clinical measures before initiating specic anti-mycobacterialtreatment maybehelpfulinthe subsequentas-sessment of treatment response. Erratic absorption of oral drugsinpatientswithCFwithpancreaticinsufciencyandpossibledrugdruginteractionsmayaffectthelevelsofthesedrugsinrespiratory secretions (122).SurgicalmanagementofNTMdiseaseinthesettingofCFmay be associated with an increased risk of mortality. Surgicalresection, lobectomy or pneumonectomy, should be reserved forthose who have an FEV1 greater than 30% predicted and havesevere, symptomatic, localizeddiseasethatfailstorespondtoaggressive medical therapy (123125). A recent study of NTMrecovered from patients with CF pre- and post-transplantationfound a strong correlation between presence of NTM pretrans-plant and recovery post-transplant. However, NTMdiseasepost-transplant was relatively uncommon, and mortality in pa-tients with NTM disease post-transplant was not different fromthose without NTM disease (105). Poor control of the mycobac-terial infectionwith medical management and, particularly, isola-tion of M. abscessus pretransplant may be risk factors for devel-opment ofpost-transplant NTM disease (105,126, 127). If thedisease ismanageable withmedical therapypretransplant, theriskandimpactofNTMrecoverypost-transplantmaybelessof a concern (105, 116, 128130).Recommendations:1. Adult and adolescent patients with CF should have peri-odic, at least yearly, screening cultures for NTM. Duringperiods of clinical decline while unresponsive to treatmentfornonNTM pathogens, allpatients with CF, includingchildren, should be evaluated for NTM (A, II).2. Patients being considered for macrolide monotherapy asimmunotherapy for CF should have sputum cultured forNTM before starting therapy and periodically thereafter,and those with repeated isolation of NTM should not re-ceive macrolide monotherapy (C, III).3. The diagnostic criteria and treatment regimens for NTMpulmonary infection in patients with CF are the same as forAmerican Thoracic Society Documents 381patients without CF, although they may be more difculttoapplybecauseofunderlyingdiseaseandconcomitantinfections (C, III).Hypersensitivity-like DiseaseAn NTM pulmonary disease syndrome with a presentation simi-lar to hypersensitivity lung disease has recently been