Macrolide-Inducible Resistance to Clindamycin and the D-Test

® CONCISE REVIEWS OF PEDIATRIC INFECTIOUS DISEASES

CONTENTS Macrolide-Inducible Resistance to Clindamycin and the D-Test

EDITORIAL BOARD Co-Editors: Margaret C. Fisher, MD, and Gary D. Overturf, MD Editors for this Issue: Michael Capello, MD, and Leonard Krilov, MD Board Members Michael Cappello, MD Geoffrey A. Weinberg, MD Charles T. Leach, MD Ellen G. Chadwick, MD Leonard Weiner, MD Kathleen McGann, MD Janet A. Englund, MD Charles R. Woods, MD Jennifer Read, MD Leonard R. Krilov, MD Jeffrey R. Starke, MD

Charles R. Woods, MD, MS

Key Words: D-test, macrolide resistance, lates resistant to clindamycin were ﬁrst rec- on”) erm expression, initial in vitro suscep- clindamycin ognized in 1968.9 Relapse of S. aureus in- tibility tests show resistance to all 3 antibi- (Pediatr Infect Dis J 2009;28: 1115–1118) fection in a rabbit model of endocarditis otic classes. When the inducible genotype

during clindamycin therapy was observed in (iMLSB) is present, in vitro tests show resis- the early 1970s.10 Clinical and bacteriologic tance to 14- and 15-membered ring macro- lindamycin has been used to treat seri- relapse in a patient with S. aureus endocar- lides listed above, but susceptibility to clin- Cous infections caused by susceptible ditis during the fourth week of clindamycin damycin and 16-membered ring macrolides Staphylococcus aureus strains in children for therapy after initial improvement was re- (available in some countries other than the 1 more than 30 years. It remains effective for ported in 1976.11 The initial isolate was US) is retained—these agents do not induce many infections caused by community-ac- susceptible to erythromycin and clindamycin resistance to themselves.12–14 quired methicillin-resistant S. aureus (CA- while that from the relapse was resistant to The inducible methylase system is 2,3 MRSA). Clindamycin also is useful for both. This led to abandonment of clindamy- regulated at the level of mRNA translation selected infections caused by pneumococci, cin for treatment of endocarditis. (translational attenuation) rather than gene group A streptococci, and a number of other The mechanism by which resistance transcription. The gene is transcribed to microbes. Absorption after oral administra- to clindamycin can emerge during therapy, mRNA constitutively, but the mRNA cannot tion is nearly complete, yielding serum con- the D-test used to detect it, and the clinical be read into protein without the presence of a centrations that approximate those from in- implications are discussed in this review. translational inducer. In the inducible erm 4 travenous (IV) administration. This permits cassette, the open reading frame of the meth- early transition to outpatient management of ylase gene and its dedicated ribosomal bind- susceptible infections without the complica- INDUCIBLE MACROLIDE 5 ing site is preceded by DNA sequences that tions of continued IV access. RESISTANCE encode another ribosomal binding site, a Clindamycin resistance is common The clinical implications of a posi- short 19 amino-acid peptide with a stop among health care-associated MRSA strains. tive D-test begin with an understanding of codon, and an additional sequence that in- Most CA-MRSA remain susceptible to date, 15–17 2 cross-resistance for 3 antibiotic families cludes 2 inverted repeats (IRs). but resistance rates vary by region. Pneu- that share a common binding site—macro- In the attenuated state, the secondary mococcal resistance to clindamycin may ex- 6 lides (eg, erythromycin, clarithromycin, structure of the transcribed mRNA has 2-stem ceed 30% in some areas of the US, while azithromcyin), lincosamides (eg, clinda- loops (Fig. 1 panel A). The ﬁrst involves the about 4% of group A streptococcal isolates 7 mycin), and group B streptogrammins (eg, last 8 codons of the short peptide (functionally, are resistant. Macrolide-inducible resistance quinupristin). This cross-resistance, called IR1) and an adjacent IR2. The second consists to clindamycin was ﬁrst recognized in the 8 the MLSB phenotype, results from enzymatic of an IR3 sequence (which follows IR2) that laboratory in the early 1960s. Clinical iso- dimethylation of an adenine residue in these binds to the sequences encoding the ribosomal antibiotics’ binding site in the 23S rRNA binding site and start codon for the methylase component of the 50S ribosomal subunit. gene (functionally, IR4). This effectively From the Department of Pediatrics, University of “hides” the ribosomal binding site and start Louisville School of Medicine, Louisville, KY. The methylase enzyme is encoded by a mul- Address for correspondence: Charles R. Woods, MD, tiallele plasmid-borne gene erm, which oc- codon for the methylase in the second (IR3– MS, Pediatric Infectious Diseases, 571 S. Floyd curs predominantly as variants erm(C) or IR4) stem-loop, such that the methylase en- St., Suite 321, Louisville, KY 40202. E-mail: erm(A) in staphylococci (other variants oc- zyme cannot be translated.15 [email protected]. When macrolide molecule capable of Copyright © 2009 by Lippincott Williams & Wilkins cur in pneumococci, hemolytic streptococci, ISSN: 0891-3668/09/2812-1115 and enterococci). When the MLSB pheno- inducing the methylase translation binds to DOI: 10.1097/INF.0b013e3181c35cc5 type is the result of constitutive (“always its target site in a ribosome, the “inhibited”

The Pediatric Infectious Disease Journal • Volume 28, Number 12, December 2009 www.pidj.com | 1115 Concise Reviews The Pediatric Infectious Disease Journal • Volume 28, Number 12, December 2009

translation—clindamycin does not induce resistance to itself, and in vitro testing of

isolates with the iMLSB genotype demon- strates clindamycin susceptibility. However, the macrolide-inducible DNA sequences that precede the erm(C) methlyase open reading frame undergo mutations, substitutions or de- letions that generate readily translatable (now

constitutive MLSB) secondary mRNA struc- tures in about 1 in 2 million replications13,14 (Fig. 1 panel C). Many infections, especially when purulent collections are present, have microbial burdens that are exceed the denom- inator of this mutation rate by 10-fold or more, such that small numbers of clindamycin-resistant microbes are likely common in infections ϩ caused by such iMLSB strains. During the past 10 years, treatment failure has reported in handfuls of adult and pediatric cases when clindamycin was used for MRSA infections caused by strains that initially appeared susceptible to clindamycin but resistant to macrolides. Clindamycin resistance was evident upon retesting of the recurring isolates.3,12,18 The common theme of these cases has been initial improvement on clindamycin with subsequent recrudescence days to weeks into or after completion FIGURE 1. A, Shows the schematic representation of the secondary structure of the of therapy. This is consistent with the above inducible methylase cassette in the absence of macrolides. Two stem loops are low frequency mutation rate from iMLS formed by the interactions of inverted repeat 1 (IR1) with IR2 and IR3 with IR4. IR4 B genotype to constitutive MLS erm expres- consists of the ribosomal binding site and initial codons of the methylase mRNA. B sion, where a resistant subpopulation sur- Ribsomes are unable to complex with the methylase binding site in this conforma- vives the innate and early adaptive host im- tion. B, Shows the secondary structure after a macrolide-bound ribosome com- mune responses and emerges to cause new or plexes with the first few codons of the short peptide and stalls. Other ribosomes worsening of existing signs and symptoms. not yet inhibited by macrolides then bind to and translate the methylase, which protects remaining ribosomes from further macrolide binding. Panel C shows an example of mutation of the “inducible” mRNA sequences that can lead to consti- THE D-TEST tutive methylase translation. The peptide and IR2 are deleted, and all mature ribo- The combination of resistance to somes are methylated. Adapted from Mol Cell. 2008;30:190–202. erythromycin with susceptibility to clindamycin in S. aureus (and other gram-positive

microbes) can be due to the iMLSB genotype ribosome is still able to bind to and start the mature ribosomes extant in the microbial or efflux pumps that remove macrolides but translation of the mRNA sequence of the cytoplasm. Most ribosomes remain macroli- not clindamycin from the microbe. The D- short peptide. The ribosome stalls after read- de-free (uninhibited) and readily able to as- test, based on disk diffusion susceptibility ing several codons but remains complexed sociate with the newly exposed erm ribo- testing, is recommended to determine if the 19 with the mRNA strand (macrolides work by somal binding site. The mRNA then is iMLSB genotype is present. In the D-test, inhibiting elongation of peptide chains be- translated, producing sufficient methylase to disks containing erythromycin (15 ␮g) and ␮ yond 3 or 4 residues). This attached stalling modify the MLSB binding sites of the ribo- clindamcyin (2 g) are placed 15 to 20 mm leads to a conformational change in the sec- somes that remain uninhibited. The microbe apart on an agar plate that has been inocu- ondary structure of the mRNA: the induc- thus is protected against further ribosomal lated with the clinical isolate (Fig. 2). A ible-state stem loops are opened up, a new inactivation despite increasing intracellular clindamycin-susceptible, erythromycin-re- single stem loop with IR2 binding to IR3 is macrolide concentrations. Phenotypic mac- sistant isolate should have a zone of inhibi- formed, and the ribosomal binding site and rolide resistance is demonstrable in vitro tion Ն21 mm in diameter, with minimal if start codon of the methylase gene are ex- within 20 to 40 minutes after initial macro- any inhibition of growth around the erythro- posed (Fig. 1 panel B).15,17 lide exposure. Of note, induction is lost after mycin disk. These initials steps occur at very low 2 generations of growth in vitro in the ab- The round zones of erythromycin and macrolide concentrations (10–100 ng/mL of sence of an inducer.15–17 clindamycin that diffuse outward from the erythromycin, below the MIC for many sus- disks partially overlap in this configuration. ceptible microbes), such as may occur Erythomycin molecules reach the outer re- shortly after a macrolide dose is adminis- RELEVANCE TO CLINDAMYCIN gion of the clindamycin zone prior to clin-

tered to a patient but before accumulation of How then is inducible macrolide re- damycin molecules. When the iMLSB geno- MIC-equivalent concentrations at the site of sistance relevant to clindamcyin? Clindamy- type is present, this leads to methylase infection. At this early stage, macrolide mol- cin binding to the same 23S rRNA subunit translation, permitting microbial growth in ecules are bound to only a small fraction of does not lead to induction of methylase this region despite subsequent diffusion of

to other antimicrobial agents based simply on the D-testϩ result in these cases may serve only to increase sensitization risks to the new agent or increase cost of therapy. Consequences of recrudescence in most initially minor episodes also are unlikely to be severe, especially if recognized early on. For sepsis, pneumonia, osteomyelitis, and other serious invasive S. aureus infections, even the small risk of emergence of resistance indicated by a positive D-test result generally should lead to avoidance of clindamycin, or prompt change from clindamycin to another agent to which the isolate is susceptible.9 When starting or continuing clindamycin for less severe infections caused by D-testϩ strains, close follow-up for potential failure and late relapse is needed. Patients and/or their caregivers should be counseled regarding potential relapse and to seek care early if they have concerns of recurrence. At the community level, relative fre-

quency of constitutive MLSB-related versus iMLSB-related clindamycin resistance should be monitored to provide appropriate guidance for empiric therapy for clinical scenarios where FIGURE 2. A, Is a photo of a positive D-test. The erythromycin and clindamycin community-acquired S. aureus infection is sus- disks are on the left and right, respectively. The expected O-shape of the clindamy- pected. Reporting aggregate clindamycin resis- cin zone of growth inhibition, which would be seen in a negative D-test, is blunted tance statistics without separating out the pro- on the side facing the erythomycin disk, resulting in a D-shaped zone. B, Is a styl- portion deemed resistant on the basis of a ized cartoon of the D-test. Erythromycin molecules diffuse into the region of the positive D-test could lead to premature aban- clindamycin zone labeled “a” before clindamycin molecules, inducing the methyl- donment of clindamycin as an empiric option ase, conferring resistance and allowing microbial growth despite subsequent arrival in clinical situations for which this may still be of otherwise inhibitory concentrations of clindamycin. Growth-inhibiting concen- an appropriate agent. The risk of failure in the trations of clindamycin reach the region labeled “b” before erythromycin mole- first hours to days of empiric clindamycin ther- cules can arrive to induce resistance. Arrows denote approximate equidistant diffu- apy prior to availability of D-test results seems sion of each antibiotic to a line bisecting the distance between the 2 antibiotic exceedingly small based on clinical reports to disks. The actual “intersection” of erythromycin versus clindamycin effects is often date—but ongoing vigilance is required. slightly bowed as in panel A. Gray areas indicate microbial growth on the agar sur- Clindamycin suppresses production of face; white areas, growth inhibition; E, erythromycin disk; C, clindamycin disk. Panton-Valentine leukocidin, alpha-hemoly- sin, and toxic-shock syndrome toxin 1 by S. aureus in vitro at the translational (ribosomal) level.21 Improved outcomes for se- inhibitory concentrations of clindamcyin. do not develop or are eliminated by host vere group A streptococcal infections treated This growth blunts the expected round zone responses before progression of infection can with clindamycin versus beta-lactam agents of growth-inhibition around the clindamycin occur. However, because of the clinical re- have been demonstrated in a mouse myositis disk into a D-shape facing the erythromycin ports of clindamycin failure associated with model22 and in a clinical case series in chil- disk (Fig. 2), which indicates a positive D- D-testϩ strains, the Clinical and Laboratory dren.23 This is the basis for use of clindamy- test. A negative D-test (round zone) indicates Standards Institute recommends that labora- cin for toxin suppression in addition to a cell efflux-mediated macrolide resistance with tories report D-testϩ isolates as resistant to wall-active agent for treatment of life-threat- retained clindamycin susceptibility. clindamycin.20 ening streptococcal and staphylococcal in- Clinical and Laboratory Standards In- fections (eg, toxic shock syndrome, necrotiz- CLINICAL IMPLICATIONS OF A stitute recommendations also suggest inclu- ing fasciitis). POSITIVE D-TEST sion of a comment that “this isolate is pre- Toxin production by isolates resistant A positive D-test indicates the pres- sumed to be resistant based on detection of to clindamcyin via constitutive methylase ence of iMLSB genotype. This means that it inducible clindamycin resistance. Clindamy- expression will not be impacted by “adjunc- is possible, but far from certain, that a sub- cin may still be effective in some patients.” tive” clindamycin. However, for infection by population of microbes resistant to clinda- This is an important allowance, as relatively strains with the iMLSB genotype, suppres- mycin may emerge and lead to clinical fail- minor infections such as cellulitis caused by sion of toxin production by clindamycin can ure or recrudescence. Infections caused by S. D-testϩ S. aureus strains (which can be reasonably be expected, at least during the aureus strains that carry the iMLSB genotype MRSA or MSSA) will respond adequately. critical first hours to days of therapy when often respond to clindamycin therapy with- These minor infections are far more common this effect is most likely to be clinically out relapse3—resistant subpopulations either than serious invasive infections, and change important. A miniscule fraction at most

(Ͻ0.01%) of the total progeny of such a acute hematogenous osteomyelitis. Pediatrics. 15. Weisblum B. Insights into erythromycin action. strain might be impervious to clindamycin 2006;117:1210–1215. Antimicrob Agents Chemother. 1995;39:797–805. early on, due to the spontaneous mutations 6. Jacobs MR, Good CE, Beall B, et al. Changes in 16. Dubnau D. Translational attenuation: the regula- that can confer constitutive expression. As serotypes and antimicrobial susceptibility of in- tion of bacterial resistance. CRC Crit Rev Bio- vasive Streptococcus pneumoniae strains in chem. 1984;16:103–132. with empiric use of clindamycin for sus- Cleveland: a quarter century of experience. J Clin pected S. aureus infections above, the deci- Microbiol. 2008;46:982–990. 17. Vazquez-Laslop N, Thum C, Mankin A. Molecular mechanism of drug-dependent ribo- sion to switch from empiric use of clindamy- 7. Richter SS, Heilmann K, Beekman S, et al. some stalling. Mol Cell. 2008;30:190–202. cin for toxin suppression to another Macrolide-resistant Streptococcus pyogenes in ribosomally active agent such as linezolid the United States, 2002–2003. Clin Infect Dis. 18. Siberry GK, Tekle T, Carroll K, et al. Failure of 2005;41:599–608. clindamycin treatment of methicillin-resistant can be based on local rates of constitutive Staphylococcus aureus expressing inducible clin- 8. Barber M, Waterworth P. Antibacterial activity of resistance among clinical isolates of S. au- damycin resistance in vitro. Clin Infect Dis. 2003; lincomycin and pristinamycin: a comparison with 37:1257–1260. reus and group A streptococci. erythromycin. Br Med J. 1964;2:603–606. REFERENCES 9. McGehee RF, Barrett F, Finland M. Resistance of 19. Steward CD, Raney P, Morrell A, et al. Testing Staphylococcus aureus to lincomycin, clinimy- for induction of clindamycin resistance in eryth- 1. Feigin RD, Pickering LK, Anderson D, et al. cin, and erythromycin. Antimicrob Agents Che- romycin-resistant isolates of Staphylococcus au- Clindamycin treatment of osteomyelitis and sep- mother. 1968;8:392–397. reus. J Clin Microbiol. 2005;43:1716–1721. tic arthritis in children. Pediatrics. 1975;55:213– 223. 10. Sande M, Johnson M. Antimicrobial therapy of 20. NCCLS (CLSI). Performance standards for anti- experimental endocarditis caused by Staphylo- microbial susceptibility testing. NCCLS (CLSI). 2. Martinez-Aguilar G, Hammerman W, Mason E coccus aureus. J Infect Dis. 1975;131:367–375. Wayne, PA, 2004. M100-S14. Jr, et al. Clindamycin treatment of invasive infections caused by community-acquired, methicillin- 11. Watanakunakorn C. Clindamycin therapy of 21. Stevens DL, Ma Y, Salmi D, et al. Impact of resistant and methicillin-susceptible Staphylococ- Staphylococcus aureus endocarditis. Am J Med. antibiotics on expression of virulence-associ- cus aureus in children. Pediatr Infect Dis J. 1976;60:419–425. ated exotoxin genes. J Infect Dis. 2007;195: 2003;22:593–598. 12. Lewis JS, Jorgensen J. Inducible clindamycin 202–211. 3. Frank AL, Marcinak J, Mangat P, et al. resistance in staphylococci. Clin Infect Dis. 2005; 22. Stevens DL, Gibbons A, Bergstrom R, et al. The Clindamycin treatment of methicillin-resistant 40:280–285. Eagle effect revisited: efﬁcacy of clindamycin, Staphylococcus aureus infections in children. Pe- 13. Daurel C, Huet C, Dhalluin A, et al. Differences erythromycin, and penicillin in the treatment of diatr Infect Dis J. 2002;21:530–534. in potential for selection of clindamycin-resistant streptococcal myositis. J Infect Dis. 1988;158: 4. Chambers HF. Protein synthesis inhibitors. In: mutants between inducible erm(A) and erm(C) 23–28. Brunton LL, ed. The Pharmacologic Basis of Staphylococcus aureus genes. J Clin Microbiol. 23. Zimbelman J, Palmer A, Todd J, et al. Improved Therapeutics. 11th ed. New York, NY: McGraw- 2008;46:546–550. outcome of clindamycin compared with beta- Hill; 2006:1188–1190. 14. Leclercq R. Mechanisms of resistance to macro- lactam antibiotic treatment for invasive Strepto- 5. Ruebner R, Keren R, Cofﬁn S, et al. Complications lides and lincosamides. Clin Infect Dis. 2002;34: coccus pyogenes infection. Pediatr Infect Dis J. of central venous catheters used for the treatment of 482–492. 1999;18:1096–1100.