The Most Frequent Aminoglycoside Resistance Mechanisms Changes with Time and Geographic Area: A Reflection of Aminoglycoside Usage Patterns?

Transcription

1 S46 The Most Frequent Aminoglycoside Resistance Mechanisms Changes with Time and Geographic Area: A Reflection of Aminoglycoside Usage Patterns? G. H. Miller, F. J. Sabatelli, R. S. Hare, Y. Glupczynski, P. Mackey, D. Shlaes, K. Shimizu, K. J. Shaw, and the Aminoglycoside Resistance Study Groups* The aminoglycoside resistance mechanisms revealed by two surveys in Europe and other countries have been compared to those revealed in earlier studies. Mechanisms have become more complex in all bacterial groups. In Providencia, Serratia, Pseudomonas, Acinetobacter, and Staphylococcus species isolates, genus-specific mechanisms were very common, and it was not possible to see differences between different geographic areas. In other Enterobacteriaceae, the increasing complexity of mechanisms was most often caused by combinations of gentamicin-modifying enzymes with AAC(6')-I, which acetylates amikacin but not gentamicin. The occurrence of these combinations varied by geographical region and among hospitals. The frequency of these combinations correlated with aminoglycoside usage in either the geographical regions or in individual hospitals. These broadspectrum combinations occurred most frequently in Citrobacter, Enterobacter, and Klebsiella species but also occurred in Escherichia, Morganella, Proteus, Salmonella, and Shigella species. Often the only clinically available aminoglycoside that retained its normal activity was isepamicin. Recently, the aminoglycoside resistance mechanisms in aminoglycoside-resistant isolates collected in a series of surveys between 1988 and 1993 were published [1, 2]. These studies showed a much greater incidence of complexity of aminoglycoside resistance mechanisms, in terms of the percentage of isolates that had more than one aminoglycoside resistance mechanism, than did earlier studies. The complexity of aminoglycoside resistance mechanisms in these studies varied by geographic region: the most complex strains were found in two regions of Latin America and in Greece and Turkey. As an example of the diversity of aminoglycoside resistance mechanisms, 53 different mechanisms were observed in the 2,080 isolates of Escherichia, Morganella, Proteus, Salmonella, and Shigella species from Europe. However, many of these aminoglycoside resistance mechanisms were relatively rare, and the 18 most common mechanisms accounted for 95.5% of the isolates. The most frequent aminoglycoside resistance mechanisms in bacteria from the European studies [2] were compared with those noted in earlier studies in the United States (USA) [3-5] and Europe [6-8]. Again, it was seen that the complexity of mechanisms had increased. This increasing complexity of aminoglycoside resistance mechanisms was suggested to be related to an increasing complexity of aminoglycoside usage. For the purpose of testing this hypothesis, aminoglycoside- * Coordinators and investigators are listed at the end of the text. Reprints or correspondence: Dr. George H. Miller, Presidential Fellow Vice President, Schering-Plough Research Institute, K15-MC4-4655, 2015 Galloping Hill Road, Kenilworth, New Jersey Clinical Infectious Diseases 1997; 24(Suppl 1):S by The University of Chicago. All rights reserved /97/ $02.00 usage information for many of the countries or regions of Europe, Latin America, and South Africa most recently surveyed [1, 2] was obtained from marketing data ( ; P. Mackey, Schering-Plough International, personal communication). In addition, two earlier surveys in the USA [4, 5] from which individual-hospital aminoglycoside-usage data were available but not previously published were included in the analysis. In order to further broaden this correlation, data from earlier published surveys in Chile and Japan [3] and from a later survey in Japan [9] were also included in this analysis. Aminoglycoside-usage data for Japan ( ) and Chile ( ) during the time of these earlier surveys were estimated by extrapolation of current usage data. Two surveys [3, 10] in the USA between 1974 and 1983 were included as well, and assumptions about aminoglycoside usage were made on the basis of historical market data. Materials and Methods The methods used to determine aminoglycoside resistance mechanisms in the most recent studies have been described previously [1, 2]. In brief, they involved the determination of an aminoglycoside resistance phenotype to 12 selected aminoglycosides by disk susceptibility testing at individual hospitals. DNA hybridization with up to 19 different aminoglycoside resistance gene probes was carried out at a central laboratory. All aminoglycoside resistance mechanisms were determined by a single person (G. H. M.) on the basis of the correlation between the two tests. Isolates for which the two tests did not agree (<1%) were excluded. Aminoglycoside resistance mechanisms revealed in the earlier studies were determined on the basis of MIC phenotypes

2 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S47 to a similar set of aminoglycosides [3], except in a study [10] that involved substrate profiles. This later method did not distinguish between AAC(6' )-I and AAC(6' )-II, and it has been assumed that all AAC(6' ) patterns were of type I in Enterobacteriaceae and type II in Pseudomonas species. All isolates from the latest American study [5] were reevaluated with 10 aminoglycoside resistance gene probes, and the aminoglycoside resistance mechanisms were adjusted to reflect the correlation of the resistance phenotype and the newly determined genotype. Later European and Japanese surveys [6-9] originally utilized a small number (3 4) of gene probes. Selected isolates from these surveys as well as from the two early studies in the USA ([3, 4], which did not use genotypes) were reanalyzed with a larger set of gene probes in order to confirm aminoglycoside resistance mechanisms. A few aminoglycoside resistance mechanisms (<1%) in isolates from each of these studies have therefore been changed from the original publications. Aminoglycoside-usage data have been obtained in a variety of ways. In the two American surveys [4, 5], individual hospital microbiologists obtained the data from their hospital pharmacies during the time of the survey. It was not possible to obtain data about prior aminoglycoside usage. Aminoglycoside-usage data for Japan were obtained from individual hospitals participating in the 1986 Japanese study [9] and also from marketshare surveys (number of units of a particular aminoglycoside sold/total aminoglycoside units sold) carried out between 1989 and These data were normalized to reflect the introduction of two new aminoglycosides, isepamicin and arbekacin, after 1986, and they agreed quite closely with those of the participating hospitals in Similar market-share data for the years between 1989 and 1993 were available for many of the countries and regions participating in the new surveys [1, 2], as well as for the USA and Japan. These yearly data were averaged for the 5-year time period, although they were quite uniform during this time in each of the surveyed regions. When internationally audited sales data were not available, we consulted individual Schering-Plough subsidiaries (in Mexico, Belgium, and South Africa) that were able to provide locally audited sales data for hospital aminoglycoside usage. In several cases (Portugal, Spain, and Latin America) in which hospital-usage data were not available, aminoglycoside pharmacy-sales data were available, and we have used these figures. It is believed that they reflect a higher use of the older, generic aminoglycosides and a lower use of amikacin than would the corresponding hospital data (personal communication, P. Mackey). 0 7', sof f ce 4'. s Z07 1 ti Z S # 4044 * sky 4 * 4) ft) 4V 00 44) 0 et) 4V 44' 0o ce,n Cb 4 4) '"') CO Nv co PS % 6) 0 Results and Discussion Aminoglycoside Resistance Mechanisms in Enterobacteriaceae The most recent studies divided Enterobacteriaceae into three groups Providencia, Serratia, and other Enterobacteria- 100 a' t Ps */ 01 Z1 01 > 01 cb 4b 0 1 cl Ps, V "1' Cb "S 01 n,' 01 Ni N. 01 S,ct, cb CO 01 0) 01 CO S. PS, 41 4, 01 Ps. 00 Figure 1. The percentages of Citrobacter, Enterobacter, and Klebsiella isolates that were resistant to clinically available aminoglycosides in various surveys of aminoglycoside resistance mechanisms [1-8] ( = gentamicin; A = tobramycin; 0 = netilmicin; 0 = amikacin; = isepamicin). ceae because the aminoglycoside resistance mechanisms in these groups were different. This occurred because chromosomal aminoglycoside-modifying enzymes have been found in Providencia stuartii [11] and Serratia marcescens [12]. The remaining Enterobacteriaceae could be divided into two additional groups based on the frequency of combinations or the complexity of the mechanisms that occurred [1]. Citrobacter, Enterobacter, and Klebsiella species had many more combinations of mechanisms (48.2%) than did Escherichia, Morganella, Proteus, Salmonella, and Shigella species (22.1%). However, the most frequent aminoglycoside resistance mechanisms observed in the two groups were very similar, and they have been combined for this study. The resistance rates among aminoglycoside-resistant Citrobacter, Enterobacter, and Klebsiella species to the clinically available aminoglycosides in these recent surveys were quite different from those observed in earlier surveys (figure 1). As expected, in the earliest surveys, rates of resistance to gentamicin and tobramycin were quite high and resistance rates to netilmicin, amikacin, and isepamicin were quite low. Later surveys in Europe, where netilmicin was frequently used, showed higher netilmicin resistance. In individual countries, the most recent surveys in Europe, Latin America, and South Africa [1] showed very high netilmicin resistance (25%-90%) and quite high rates of amikacin resistance (14%-92%). While figure 1 shows the data for Citrobacter, Enterobacter, and Klebsiella species, among which this trend was most pronounced, a qualitatively similar trend was observed in other gram-negative bacteria and staphylococci..., ti

3 S48 Miller et al. CID 1997;24 (Suppl 1) The high rates of resistance to netilmicin and amikacin in Enterobacteriaceae were due to combinations of previously common gentamicin-modifying enzymes with AAC(6' )-I, an enzyme that confers resistance to tobramycin, netilmicin, amikacin, dibekacin, and kanamycin but not to gentamicin and isepamicin (hereafter referred to as AAC(6' )-I [tobramycin, netilmicin, amikacin, dibekacin, and kanamycin]; other mechanisms are referred to with similar constructions). While this enzyme occurred in 11.7% of the aminoglycoside-resistant isolates in the most recent surveys as a single mechanism, it occurred in 34.0% of the isolates in combination with other aminoglycoside resistance mechanisms. This observation suggested that these combinations had occurred as a result of a changed aminoglycoside selective pressure in these regions. To examine this question, we have attempted to correlate aminoglycoside resistance mechanisms (noted in earlier studies as well as the current studies) with known or estimated aminoglycoside usage. The most frequent aminoglycoside resistance mechanisms in the three earliest surveys are shown in figure 2. As can be seen, the principal mechanism in the USA isolates was ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin). While this mechanism was also common in Japan, AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) was found alone (12.8%) and combined with ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) (37.6%). This difference in aminoglycoside resistance mechanisms between the American and Japanese surveys accounts for the eightfold higher amikacin resistance rate in the Japanese isolates (figure 1). Similar aminoglycoside resistance mechanisms (figure 2) were found in a more recent (1986) survey of nine Japanese hospitals [9]. Aminoglycoside usage in four of the hospitals surveyed was as follows: gentamicin, 5%-25%; tobramycin and dibekacin, 20%-50%; and amikacin, 30%-50%. These figures do not include oral kanamycin usage, which was 150%-249% higher than the parenteral usage of all the other aminoglycosides in three of the hospitals. These data are quite similar to market-share data for Japan in 1989 (figure 3; tobramycin plus dibekacin, 55%, and amikacin, 30%). This contrasts quite sharply with the 1989 data for the USA shown in figure 3 (gentamicin, 80%; tobramycin, 10%; and amikacin, 10%). While exact aminoglycoside-usage data for the USA at the time of these early surveys were not available, it is known that gentamicin was the most frequently used aminoglycoside, and we believe that the 1989 figures would not be very different from those for (except that tobramycin usage was probably higher). During the earliest studies in the USA ( ), gentamicin usage was probably even higher and that of amikacin was probably lower. Since ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) was found in both countries and AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) was found almost exclusively in Japan, a correla- Figure 2. Comparison of the 14 most frequent aminoglycoside resistance mechanisms in early studies of Enterobacteriaceae (other than Providencia and Serratia species) in the USA (A and B) [3, 10] and in Japan (C and D) [3, 9]. The numbers above each bar are percentages of the total no. of resistant isolates in a given study that had that particular mechanism. The phenotype of each mechanism for the clinically available aminoglycosides is shown after the mechanism (e.g., GTNAI = resistance to gentamicin, tobramycin, netilmicin, amikacin, and isepamicin). Per investigation, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (E) were as follows: A [3], n = 415, E = 99.0%; B [10], n = 51, E = 92.2%; C [3], n = 282, E = 98.9%; and D [9], n = 116, E = 99.2%. tion between observed aminoglycoside resistance mechanisms and usage was implied [3, 9]. In 1984, small numbers of aminoglycoside-resistant isolates from individual hospitals in the USA and Chile were tested in our laboratories (by R. S. H.), and these isolates had combinations of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) with various gentamicin-modifying enzymes. At the same time, Levine et al. [13] reported similar aminoglycoside resistance mechanisms in isolates in a New York City hospital and suggested a correlation with tobramycin and amikacin usage. Two surveys of aminoglycoside resistance mechanisms were carried out in the USA during and

4 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S49 Figure 3. Estimated market-share data for aminoglycosides between 1989 and The estimated usage of netilmicin (N), tobramycin (T), and dibekacin (D) is plotted versus amikacin (A) usage in terms of percentage of market share. Gentamicin (G) usage is estimated by subtraction, e.g., French hospitals utilized an average of 39% netilmicin and tobramycin and 26% amikacin, so gentamicin usage was 35%. French hospitals were classified as belonging to the GNA group since the actual percentage of tobramycin and dibekacin use was small (8.8%). The hospitals belonging to each usage group have been indicated by the symbols in the enclosed areas on the graph: e.g., GN = gentamicin/netilmicin usage group. Hospital-usage data for South Africa and Latin America are unavailable, but it has been suggested that such hospitals would be in the GA group on the basis of pharmacy-use data (shown) and market data (P. Mackey, Schering-Plough International, personal communication). cl et xi =,44 E V GN OD GTN GT 10 - G GA? Estimated use of amikacin (%) European Hospitals German Hospitals Italian Hospitals Portuguese Pharmacies Spanish Pharmacies * French Hospitals * Greek Hospitals + Belgian Hospitals Latin American 0 Pharmacies Mexican Hospitals South African Pharmacies U.S. Hospitals O Japanese Hospitals [4, 5]; individual-hospital aminoglycoside-usage data were collected at the time of the surveys. The most frequent aminoglycoside resistance mechanisms in these two surveys are shown in figures 4 and 5, separated into aminoglycoside-usage groups. The aminoglycoside-usage groups were determined on the basis of the graphs shown in figures 6 and 7. The most frequent aminoglycoside resistance mechanisms in the first study (figure 4), noted in the gentamicin and gentamicin/tobramycin usage groups, were similar to those in earlier studies; there were very high (54.6%-66.7%) levels of ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin). However, several combinations of gentamicin-modifying enzymes with AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) were found (5.8%-8.7%). In contrast, in the gentamicin/tobramycin/amikacin and gentamicin/amikacin usage groups, the incidence of gentamicin-modifying enzymes was significantly reduced (3.2%-39.3%) and the incidence of combinations of gentamicin-modifying enzymes with AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) was quite high (4.8%-29.1%). The incidence of AAC(6' ) -I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) alone was also significantly greater (7.1%-8.1%) than in the two other usage groups (zero). The most frequently occurring combination in these two groups, AAC(3)-II (gentamicin, tobramycin, and netilmicin) and AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin), was the same as that which occurred in the earlier outbreak in New York City [13]. In the second study (figure 5), aminoglycoside usage was found to have changed. Many of the previously gentamicin/ tobramycin-using hospitals became either gentamicin users or gentamicin/tobramycin/amikacin users. A small group of hospitals using a very small percentage of gentamicin (netilmicin/ tobramycin/amikacin users) were also included. However, the most frequent aminoglycoside resistance mechanisms were similar to those in the first study, i.e., the gentamicin- and gentamicin/tobramycin-usage groups had a high incidence of gentamicin-, tobramycin-, and netilmicin-modifying enzymes and a lower incidence of combinations that included AAC(6' )- I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) (1.3%-8.0%). These combinations were much more frequent in the remaining groups, which utilized amikacin (0.9%-15.6%). For the first time, significant proportions (4.3% 18.9%) of a relatively new [14] gentamicin-modifying enzyme, AAC(3)-VI (gentamicin, tobramycin, and netilmicin), and permeability resistance (all aminoglycosides) were observed. These aminoglycoside resistance mechanisms accounted for the decreased incidence of ANT(2" )-I (gentamicin, tobramycin, dibekacin, and kanamycin) in the gentamicin- and gentamicin/tobramycin-using hospitals, while combinations that included AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) accounted for its decrease in the other hospitals. The earliest surveys of aminoglycoside resistance mechanisms in Europe and Latin America [3, 6 8] show a similar trend with regard to time (figure 8). The incidence of gentami-

5 S50 Miller et al. CID 1997;24 (Suppl 1) amikacin, dibekacin, and kanamycin), but the numbers of aminoglycoside-resistant isolates tested were relatively small. It is noteworthy, however, that the largest number of these combinations was found in the Belgian survey [8], where the use of amikacin was very high (figure 3). The most recent surveys (figure 9) were carried out to determine if this trend was correct. As can be seen in figure 9, combinations of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) with AAC(3)-II (gentamicin, tobramycin, and netilmicin) were as high as 8.3% in Europe and 32.2% in Latin America/South Africa. Other broad-spectrum combinations, seen previously, were also common. In Latin America, particularly in Venezuela, a new phenotype of Figure 4. Comparison of the 14 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in a single study in the USA [4], in which the participating hospitals were classified into four aminoglycoside-usage groups (see figure 6 for definitions). The graphs are organized as described in figure 2; per group, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (1) were as follows: A, n = 69, E = 100.0%; B, n = 119, E = 97.6%; C, n = 28, E = 100.0%; and D, n = 62,E = 98.4%. cin-modifying enzymes was very high in Europe [6] and Chile [3] and similar to that in the USA except that the most common gentamicin-modifying enzyme was AAC(3)-II (gentamicin, tobramycin, and netilmicin) and the incidence of ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) was lower. The very high incidence of AAC(3)-II (gentamicin, tobramycin, and netilmicin) in Chile explains the very high netilmicin resistance seen in figure 1. This is consistent with results in an individual hospital in England [15] and one in Spain [16] and may be due to geographic factors. However, it should be noted that netilmicin usage in Europe and Latin America was substantial (20%-50%) at the time of these surveys. The more recent surveys in Europe [7, 8] (figure 8) show an increased incidence of combinations of gentamicinmodifying enzymes with AAC(6' )-I (tobramycin, netilmicin, Figure 5. Comparison of the 14 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in a single study in the USA [5], in which the participating hospitals were classified into five aminoglycoside-usage groups (see figure 7 for definitions). The graphs are organized as described in figure 4; per group, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (E) were as follows: A, n = 150, = 98.0%; B, n = 37, E = 97.3%; C, n = 32, E = 100.0%; D, n = 235, E = 96.2%; and E, n = 31, E = 96.8%.

6 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S a E N GT k CA I* GA 0A Estimated use of amikacin (%) Figure 6. Aminoglycoside usage in 26 hospitals in the USA at the time of their participation in a survey of aminoglycoside resistance mechanisms ( ) [4]. The graph is similar to figure 3 except that each point is an individual hospital and that there was little or no usage of netilmicin or dibekacin in these hospitals. The groupings were based on similarity of usage. In the few cases where individual hospitals could have been placed in neighboring groups, the assignment was based on similarity of mechanisms (A = hospitals using gentamicin [G]; = hospitals using gentamicin/tobramycin [GT]); = hospitals using gentamicin/tobramycin/amikacin [GTA]; = hospitals using gentamicin/amikacin [GA]). amikacin and isepamicin resistance was seen. This new phenotype, "?R?," was never observed alone and could represent an error or artifact in aminoglycoside resistance mechanism assignment. (The cause of?r? has been identified in individual isolates from Venezuela, Belgium, and Turkey as a new type of AAC( 6' ), which we propose to call AAC( 6' )-III. It has the same resistance phenotype as AAC( 6' )-I [tobramycin, netilmicin, amikacin, dibekacin, and kanamycin] except that it confers an unusually high level of resistance to amikacin and, unlike AAC(6' )-I, confers resistance to isepamicin. Three new genes that confer this resistance pattern have been cloned. Ongoing hybridization studies have shown that most but not all isolates with the?r? phenotype have one of these three genes.) However, resistance to both amikacin and isepamicin was confirmed in these isolates, a phenomenon that is not explainable by the currently cloned aminoglycoside resistance genes [14]. Figure 3 shows the aminoglycoside-usage data for hospitals in 14 European countries: tobramycin and netilmicin, 50%; amikacin, 7%; and gentamicin, 40%. Similar data were not available for hospitals in Latin America. However, outpatient use of amikacin from pharmacies in Latin America, Mexico, and South Africa was 10%-20%, and on the basis of parallel extrapolation to hospital usage in Mexico and South Africa, it was estimated that hospital usage of amikacin could range from 30% to 60%. Thus, the higher incidence of combinations of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) as well as the "?R?" phenotype in Latin America and South Africa could be correlated with higher amikacin usage. The incidence of broad-spectrum combinations of aminoglycoside resistance mechanisms in Europe was not evenly distributed among the various countries and regions that were surveyed. Figure 10 shows the most frequent aminoglycoside resistance mechanisms in two large surveys. The isolates studied were from Germany, Italy/Portugal/Spain, and several northern and central European countries (103 of the 168 isolates in this latter group were from two hospitals in the United Kingdom). Aminoglycoside usage in German hospitals (figure 3) was 80% gentamicin use and 20% tobramycin and netilmicin use. The most common aminoglycoside resistance mechanisms were ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) and AAC(3)-II (gentamicin, tobramycin, and netilmicin), as well as an unusual aminoglycoside resistance mechanism, AAC(3)-IV (gentamicin, tobramycin, netilmicin, and apramycin). This later enzyme was quite common (10.9% overall) in Escherichia coli isolated in all surveyed countries CcZ z a = 60- T a E ca.0 e ea E T., J. siw Estimated use of amikacin (%) Figure 7. Aminoglycoside usage in 31 hospitals in the USA at the time of their participation in a survey of aminoglycoside resistance mechanisms ( ) [5]. The graph is similar to figure 6 except that one hospital in the NTA group used netilmicin (70%), while the two other hospitals in this group used tobramycin (A = hospitals using gentamicin [G]; = hospitals using gentamicin/tobramycin [GT]; = hospitals using gentamicin/amikacin [GA]; = hospitals using gentamicin/tobramycin/amikacin [GTA]; = hospitals using netilmicin/tobramycin/amikacin [NTA]).

7 S52 Miller et al. CID 1997;24 (Suppl 1) Figure 8. Comparison of the 14 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in studies in Chile [3], Europe [6, 7], and Belgium [8]. The graphs are organized as described in figure 2; per investigation, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (E) were as follows: A [3], n = 263, E = 99.6%; B [6], n = 122, E = 98.4%; C [7], n = 51, E = 98.0%; and D [8], n = 98, E = 98.0%. in which apramycin was used in animal husbandry (Belgium, England, France, and Germany) but almost absent where apramycin was apparently not used (Italy/Portugal/Spain and Greece/Turkey). Amikacin usage in Italian hospitals was 15.8%, and among outpatients in Portugal and Spain it was 3.2% and 5.6%, respectively (figure 3). Hospital usage of amikacin in Portugal and Spain is thought to be similar to that in Italy, thus placing these three countries in a gentamicin/tobramycin/netilmicin-usage group. AAC(3)-II (gentamicin, tobramycin, and netilmicin) was responsible for aminoglycoside resistance in more than 60% of the isolates from this region; only 7.6% of the isolates had a combination of this enzyme with AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin). The low incidence of broad-spectrum combinations in Germany and Italy/Portugal/Spain contrasts markedly with that seen in the three other European surveys (figure 11). In Belgian hospitals, where amikacin and netilmicin usage (figure 3; 60% and 30%, respectively) was very high, the incidence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) as a single aminoglycoside resistance mechanism was high (20.3%), and that of broad-spectrum combinations was also high (>18.4%). In French hospitals, where there was almost equal usage of gentamicin, netilmicin, and amikacin (figure 3; 29%, 30%, and 26%, respectively), the incidence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) as a single enzyme was quite high (18.3%), and the incidence of this enzyme in combinations was >27.2%. In Greek hospitals aminoglycoside usage was similar to that in Belgian hospitals (figure 3; =40% amikacin use and 50% netilmicin use). The occurrence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) as a single enzyme was lower (8.9%), but a single combination of AAC(6' )-I with AAC(3)-I (gentamicin) was very common (59.5%), especially in the 14 Greek and one Turkish hospital surveyed. This combination occurred in many different types of Enterobacteriaceae in each of the individual Greek hospitals, where it seemed to be acting as a plasmid epidemic. It was almost always found together with APH(3' )-I (kanamycin and neomycin). Thus, Belgium, France, and Greece all had a much higher incidence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin), both alone and in combinations, than did the other European countries, and this incidence correlated with a much higher (>25%) usage of amikacin. Figure 12 shows the data from the individual Latin American and South African surveys as well as data from four hospitals in Asia. Aminoglycoside resistance mechanisms in Argentina, Chile, and Uruguay were quite similar to those seen in Guatemala, Mexico, and Venezuela, except that the incidence of the unknown phenotype "?R?" was much higher in the four Venezuelan hospitals. Among the first three countries, this phenotype was almost exclusively found in Proteus mirabilis isolates from Chile. While exact aminoglycoside-usage data from hospitals in Latin America (other than Mexico) were not available, it is known that amikacin usage there was quite high (20%-50%) and probably similar to that in Mexico (figure 3). The aminoglycoside resistance mechanism data from the three South African hospitals surveyed showed an unusually high incidence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin), alone as well as in combination. This correlates with both amikacin-usage data (figure 3) and a known outbreak of resistant Klebsiella species in one of the hospitals [17]. The four Asian hospitals were a very heterogeneous group in terms of observed aminoglycoside resistance mechanisms and probably also in terms of aminoglycoside usage. Almost all of the combinations of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) occurred in one hospital in the Philippines, where amikacin usage was probably high.

8 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S53 Figure 9. Comparison of the 18 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in recent ( ) studies in Europe [1, 2], and Latin America/South Africa [1]. The graphs are organized as described in figure 2 except that it was necessary to include four additional mechanisms (see page S51 for explanation of "?R?"); per investigation, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (E) were as follows: A [1, 2], n = 3,583, E = 95.5%; and B [1], n = 1,526, E = 94.0%. Aminoglycoside Resistance Mechanisms in Providencia Species The chromosomal aminoglycoside-modifying enzyme from P. stuartii, mentioned above [11], is AAC(2' )-I (gentamicin, tobramycin, netilmicin, dibekacin, and neomycin). As can be seen in table 1, this enzyme was the most common mechanism in Providencia species. In the earliest studies in the USA [3, 10] and Europe [6], this aminoglycoside resistance mechanism accounted for more than 80% of the isolates. In later studies in the USA [4, 5] and Japan [3, 9], other aminoglycoside resistance mechanisms were observed. This trend continued in the more recent European [1, 2, 8] and Latin American/South African surveys [1]. Although the total number of aminoglycoside-resistant isolates studied in the recent surveys was large, when these numbers were divided according to countries or regions of origin (other than Europe or Latin America/South Africa), they became too small to allow comparison with aminoglycoside-usage patterns. Several regional differences did exist, however. Isolates from the USA had a high incidence of AAC(2' )-I (gentamicin, tobramycin, netilmicin, dibekacin, and neomycin), alone and in combination with ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin). Isolates from Chile [3] and in the recent Latin American studies [1] had an unusually high incidence of AAC(3)-II (gentamicin, tobramycin, and netilmicin) and a lower incidence of resistance due to the chromosomal aminoglycoside resistance mechanism. In contrast, the isolates from Japan [3, 9] had a high incidence of the combination of ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) and AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin). The most recent European studies [1, 2, 8] found a high incidence of AAC(2' )-I (gentamicin, tobramycin, netilmicin, dibekacin, and neomycin), as was found in the USA, but combinations with other aminoglycoside resistance mechanisms were also seen. It was not possible to relate these regional and temporal differences to differences in aminoglycoside usage because of the small number of Providencia species studied. These differences are similar, however, to those described for other Enterobacteriaceae. Aminoglycoside Resistance Mechanisms in Serratia Species A chromosomal aminoglycoside-modifying enzyme, AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and

9 S54 Miller et al. CID 1997;24 (Suppl 1) Figure 10. Comparison of the 18 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in recent ( ) studies in Europe [1, 2]. The isolates have been classified according to their geographic origin: 15 hospitals in Germany; 8 hospitals in northern and central European countries; 9 hospitals in Italy; 1 in Portugal; and 5 in Spain. The graphs are organized as in figure 9; per investigation, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (/) were as follows: A [1, 2], n = 308, I = 94.2%; B [1, 2], n = 168, / = 92.1%; and C [1, 2], n = 657, I = 96.7%. kanamycin), was present in S. marcescens. Table 2 shows that this enzyme remained the single most common aminoglycoside-modifying enzyme in Serratia species. In studies in the USA and Japan, it was found alone and in combination with ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin). Since this aminoglycoside resistance mechanism was common in Enterobacteriaceae in both the USA and Japan, this is not surprising. In Europe, Latin America, and South Africa, where AAC(3)-II (gentamicin, tobramycin, and netilmicin) was a very common aminoglycoside resistance mechanism in Enterobacteriaceae, AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) was often found combined with this enzyme as well as ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin). Thus, aminoglycoside resistance mechanisms in Serratia species comprise the chromosomal aminoglycoside resistance mechanism and the common plasmid-mediated aminoglycoside resistance mechanisms of Enterobacteriaceae. Consistent with this generalization is the recent finding [1, 2] that 15% of Serratia species expressed a plasmid-mediated AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) as well as the chromosomal type of resistance mechanism. Thus, these Serratia isolates had two genes, aac(6' )-Ib and aac(6' )- Ic encoding enzymes, which were capable of causing the observed resistance. Because of the predominance of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) in Serratia species, it was not possible to see any differences in aminoglycoside resistance mechanisms associated with aminoglycoside usage in the early or current studies. However, as in other Enterobacteriaceae, aminoglycoside resis-

10 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S55 Figure 11. Comparison of the 18 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in recent ( ) studies in Europe [1, 2]. The isolates have been classified according to their geographic origin: A, 12 hospitals in Belgium, 1 hospital in Luxembourg, and 1 in the Netherlands; B, 55 hospitals in France; and C, 13 hospitals in Greece and 1 hospital in Turkey. The graphs are organized as in figure 9; per investigation, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (E) were as follows: A [1, 2], n = 163, E = 98.2%; B [1, 2], n = 2,018, E = 96.2%; and C [1, 2], n = 269, 1 = 93.3%. tance mechanisms in these isolates have become more complex as the usage of aminoglycosides has increased. Aminoglycoside Resistance Mechanisms in Pseudomonas Species The most common aminoglycoside resistance mechanisms in Pseudomonas species have always been different from those in Enterobacteriaceae (table 3). Three of the four common plasmid-mediated enzymes in Enterobacteriaceae, AAC(3)-I (gentamicin), AAC(3)-II (gentamicin, tobramycin, and netilmicin) and AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin), were rare. Two different aminoglycoside resistance mechanisms, AAC(6' )- II (gentamicin, tobramycin, and netilmicin) and permeability resistance (all aminoglycosides), were common, as was one of the Enterobacteriaceae aminoglycoside resistance mechanisms, ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin). The one exception among the surveys shown in table 3 is the relatively high incidence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) in Japan, where this enzyme was very common in Enterobacteriaceae (figure 2). Although many additional aminoglycoside resistance mechanisms were seen in the latest two surveys in the USA, no association with the aminoglycoside-usage groups of the hospitals was seen (data not shown). This is partly because the number of isolates per group was small but also because permeability resistance (all aminoglycosides) was found at similar frequencies in all groups, which may have obscured any differences in enzyme distribution.

11 S56 Miller et al. CID 1997;24 (Suppl 1) Figure 12. Comparison of the 18 most frequent aminoglycoside resistance mechanisms in Enterobacteriaceae (other than Providencia and Serratia species) in recent ( ) non-european studies [1]. The isolates have been classified according to their geographic origin: 4 hospitals in the Pan-Pacific region; 3 hospitals in South Africa; 9 hospitals in Argentina/Chile/Uruguay; and 6 hospitals in Guatemala/Mexico/Venezuela. The graphs are organized as in figure 9; per investigation, the total number of isolates studied and the sum of the percentages of all isolates that had one of the most frequent mechanisms (E) were as follows: A [1], n = 394, E = 97.7%; B [1], n = 504, E = 94.4%; C [1], n = 740, E = 97.0%; and D [1], n = 282, E = 85.1%. Table 3 shows a distribution of aminoglycoside resistance mechanisms in European hospitals between 1984 and 1988 that is quite similar to that seen in the USA. The Chilean hospitals, however, were quite different in that a very high percentage of the isolates had AAC(3)-I (gentamicin) and AAC(3)-II (gentamicin, tobramycin, and netilmicin). This latter aminoglycoside resistance mechanism was, of course, very common in Enterobacteriaceae in Chile at that time (figure 8 and tables 1 and 2). Thus, the Chilean and Japanese isolates demonstrated that all of the common Enterobacteriaceae aminoglycoside resistance mechanisms can occur in Pseudomonas isolates; however, only ANT(2" )-I (gentamicin, tobramycin, dibekacin, and kanamycin) was common in the USA and Europe. The most recent European and Latin American studies are also shown in table 3. The three aminoglycoside resistance mechanisms common in Pseudomonas species ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin), AAC(6' )-II (gentamicin, tobramycin, and netilmicin), and permeability resistance (all aminoglycosides) were still cornmon, but a very large number of different combinations were seen in these studies. The most common of these combinations involved the addition of permeability changes to the other two common Pseudomonas inactivating enzymes. Combinations of permeability resistance (all aminoglycosides) with ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) or AAC(6' )-II (gentamicin, tobramycin, and netilmicin) have increased significantly from previous studies; how-

12 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S57 Table 1. Comparison of the most frequent aminoglycoside resistance mechanisms in Providencia species. Data from indicated surveys Latin America/ Location USA USA Chile Japan USA Europe USA Belgium Europe South Africa Period Reference(s) [3] [10] [3 ] [3, 9] [4] [6] [5] [8] [1, 2] [1] Isolates (n) Total percentage of aminoglycoside resistance Aminoglycoside resistance mechanism(s) Phenotype* Percentage of Providencia isolates with aminoglycoside resistance mechanism ANT(2")-I GT AAC(3)-II GT AAC(2')-I GTN AAC(2')-I + ANT(2")-I GTN AAC(2')-I + AAC(3)-I GTN AAC(2')-I + AAC(3)-II GTN AAC(6')-I TNA AAC(2')-I + AAC(6')-I GTNA ANT(2")-I + AAC(6')-I GTNA AAC(3)-II + AAC(6')-I GTNA AAC(2')-I + Permeability GTNAI * A = amikacin; G = gentamicin; I = isepamicin; N = netilmicin; T = tobramycin. ever, it was not possible to relate this to any specific changes in aminoglycoside usage, since they occurred at similar frequencies in most surveys. In fact, it may be partly due to methodological changes, since a more uniform definition of permeability resistance was used [1, 2] than in previous surveys and the use of genotyping made the detection of these combinations easier. While the five most common aminoglycoside resistance mechanisms accounted for 74.9% of all Pseudomonas isolates tested, the remaining isolates had 80 different aminoglycoside resistance mechanisms. Most of the other aminoglycoside resistance mechanisms occurred in <1% of the total isolates and seemed to be associated with individual hospitals rather than a country or region (data not shown). Thus, unlike those in Enterobacteriaceae, aminoglycoside resistance mechanisms in Pseudomonas species do not seem to be related to aminoglycoside usage, other than the fact that over time these mechanisms have become more complex. Aminoglycoside Resistance Mechanisms in Acinetobacter and Staphylococcus Species Data from the most recent European and Latin American surveys of Acinetobacter and Staphylococcus species have been published [1]. Early surveys of aminoglycoside resistance mechanisms either did not include Acinetobacter or Staphylococcus species or the number of isolates tested was small. Therefore, it was not possible to relate the current aminoglycoside resistance mechanisms in these genera to earlier data. However, like those in Pseudomonas species, aminoglycoside resistance mechanisms in Acinetobacter species were very complex; 67 aminoglycoside resistance mechanisms were noted in 1,189 isolates. Because of the occurrence of two aminoglycoside resistance mechanisms, APH(3')-VI (amikacin, isepamicin, kanamycin, and neomycin) and AAC(3)-? (gentamicin and netilmicin), at very high frequencies (47.6% and 51.4%, respectively), it was not possible to see consistent differences in aminoglycoside resistance mechanisms between surveys (data not shown). Thus, no relationship to aminoglycoside usage was evident. Unlike those in Acinetobacter species, aminoglycoside resistance mechanisms in Staphylococcus species [1] were quite simple; three single aminoglycoside resistance mechanisms were found alone and in all possible combinations. In fact, 49.1% of the isolates had multiple aminoglycoside resistance mechanisms. The bifunctional aminoglycoside resistance mechanism APH(2") + AAC(6' ) (gentamicin, tobramycin, netilmicin, amikacin, isepamicin, dibekacin, kanamycin, fortimicin, and arbekacin) was the

13 S58 Miller et al. CID 1997;24 (Suppl 1) Table 2. Comparison of the most frequent aminoglycoside resistance mechanisms in Serratia species. Data from indicated survey Location Period Reference(s) Isolates (n) Total percentage of aminoglycoside resistance USA USA Chile Japan [3 ] [10] [3 ] [3] USA [4] Europe Japan [6] [9] USA Europe [5 ] [7] Belgium Europe Latin America/ South Africa [8] [1, 2] [1] Aminoglycoside resistance mechanism(s) Phenotype* Percentage of Serratia isolates with aminoglycoside resistance mechanism AAC(3)-I G ANT(2")-I GT AAC(3)-II GTN AAC(6')-I INA AAC(6')-I + AAC(3)-II GTNA AAC(6')-I + AAC(3)-II + AAC(3)-I GTNA AAC(6')-I + AAC(3)-I GTNA AAC(6')-I + ANT(2")-I GTNA * See footnote to table 1 for definitions. most common (90.8%) and occurred at similar frequencies in all surveys. The other aminoglycoside resistance mechanisms also occurred at similar frequencies in all surveys. Thus, again, no relationship to aminoglycoside usage was observable. Conclusions Aminoglycoside resistance mechanisms in all types of bacteria have become more complex with the increased usage of aminoglycosides over time. Combinations of mechanisms have occurred that have broadened the spectrum of aminoglycoside resistance in all genera. In Providencia and Serratia species in which chromosomal aminoglycoside resistance mechanisms are present, this complexity has consisted of the addition of plasmid-mediated aminoglycoside resistance mechanisms. In Pseudomonas species, combinations of plasmid-mediated aminoglycoside resistance mechanisms and permeability changes have led to a broader spectrum of resistance. Although data from earlier surveys are lacking, similar changes seem to have taken place in Acinetobacter and Staphylococcus species. The changes in aminoglycoside resistance mechanisms seen in these five genera seem to have occurred to a similar extent in all geographic regions surveyed and thus do not seem to be related to specific aminoglycoside usage; rather, these changes seem to be a response to continued usage of aminoglycosides in general. Unlike the changes in aminoglycoside resistance mechanisms in the five genera mentioned above, the increasing complexity of aminoglycoside resistance mechanisms in other Enterobacteriaceae do vary within different geographic regions and hospitals. The increasing complexity of aminoglycoside resistance mechanisms was found to be greatest in Citrobacter, Enterobacter, and Klebsiella species but was also seen in Escherichia, Morganella, Proteus, and Salmonella species. The combinations of aminoglycoside resistance mechanisms that occurred most frequently were gentamicin-modifying enzymes that were common in early surveys, combined with AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin). The gentamicin-modifying enzymes that were common in specific geographic areas in early surveys remained common in later surveys, i.e., ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) in the USA and Japan and AAC(3)-II (gentamicin, tobramycin, and netilmicin) in Europe and Latin America. Thus, the most frequent combinations of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) were with ANT(2")-I (gentamicin, tobramycin, dibekacin, and kanamycin) in the USA and Japan, while those in Europe and Latin America were with AAC(3)-II (gentamicin, tobramycin, and netilmicin). Combinations with AAC(3)-I (gentamicin) were also common in Greece and France [1, 2]. Regardless of the individual combination that occurred in a given geographic area, the effect on aminoglycoside resistance

14 CID 1997;24 (Suppl 1) Use of and Resistance to Aminoglycosides S59 Table 3. Comparison of the most frequent aminoglycoside resistance mechanisms in Pseudomonas species. Data from indicated survey Latin America/ Location USA USA Chile Japan USA Europe Japan USA Europe Belgium Europe South Africa Period Reference(s) [3] [10] [3] [3] [4] [6] [9] [5] [7] [8] [1, 2] [1] Isolates (n) , Total percentage of aminoglycoside resistance Aminoglycoside resistance mechanism(s) Phenotype* Percentage of Pseudomonas isolates with aminoglycoside resistance mechanism AAC(3)-I G ANT(2")-I GT AAC(3)-III GN AAC(3)-? GN AAC(3)-III + ANT(2")-I GTN 1.2 AAC(6')-II GTN AAC(6')-II + ANT(2")-I GTN AAC(6')-II + AAC(3)-I GTN AAC(3)-II GTN AAC(3)-VI GTN AAC(3)-III + APH(3')-VI GNAI 1.2 AAC(6')-I TNA AAC(6')-I + AAC(3)-I GTNA AAC(6')-I + ANT(2")-I GTNA AAC(6')-II + APH(3')-VI GTNAI 4.1 AAC(6')-II + APH(3')-VI + AAC(3)-I GTNAI 3.6 AAC(6')-II + ANT(2")-I +? R? GTNAI AAC(6')-II + AAC(3)-II +? R? GTNAI AAC(6')-II + AAC(3)-II +? R? + AAC(3)-I GTNAI Permeability GTNAI Permeability + AAC(6')-II GTNAI Permeability + AAC(6')-II + AAC(3)-II +? R? GTNAI Permeability + AAC(6')-II + AAC(3)-I GTNAI Permeability + ANT(2")-I GTNAI Permeability + AAC(3)-I GTNAI 1.1 * See footnote to table 1 for definitions.

15 S60 Miller et al. CID 1997;24 (Suppl 1) was the same: broad-spectrum resistance to gentamicin, tobramycin, netilmicin, and amikacin. Among the clinically available aminoglycosides tested, only isepamicin consistently retained activity against these isolates (figure 1) [2]. While these broad-spectrum combinations occurred in all of the surveys initiated after 1984, they occurred more frequently in some geographic areas or hospitals than in others. The two studies in the USA show that they occurred more often in hospitals where aminoglycoside usage was more complex than administration of gentamicin alone or gentamicin plus tobramycin [4, 5]. The two surveys in Japan [3, 9], where parenteral aminoglycoside usage was mostly with tobramycin, dibekacin, and amikacin, revealed a much higher incidence of these combinations than were seen in the USA, where overall aminoglycoside usage was and remains mostly administration of gentamicin (-80%). In the most recent studies [1, 2], these combinations were unusual in Germany and Italy/Portugal/Spain, where aminoglycoside usage was either with gentamicin/tobramycin or with gentamicin/tobramycin/netilmicin. In contrast, in Belgium, France, and Greece, where amikacin usage was >25%, these combinations are very common. Similar aminoglycoside resistance mechanisms were also seen in Latin America and South Africa, where it is believed that amikacin usage is also >25%. Thus, in Enterobacteriaceae in which plasmid-mediated aminoglycoside resistance mechanisms predominate, there is a correlation between the type of aminoglycoside resistance mechanism combinations that are most frequent in a given geographic area or hospital and aminoglycoside selective pressure. While this paper suggests a correlation between aminoglycoside usage and mechanisms of resistance, it does not address the question of the frequency of resistance to aminoglycosides. The results of a survey of aminoglycoside resistance in 12 different countries, carried out in 1986 [18], suggested a correlation between total aminoglycoside consumption and gentamicin resistance. A publication in 1995 [19] provides the opportunity to compare the relationship of the aminoglycoside-usage groups in this article with aminoglycoside resistance rates in five European countries. Table 4 compares the data from this article with resistance data extracted from the 1995 survey [19]. It can be seen that in West Germany and Italy, where the incidence of AAC(6' )- I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin) alone and in combinations was similar and correlated with their aminoglycoside usage groups, the incidence of resistance in Enterobacteriaceae was quite different. The rates of resistance are higher in Italy than in West Germany. If one compares Belgium and Greece, which have similar aminoglycoside-usage patterns and similarly high incidences of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin), the rates of resistance in Enterobacteriaceae are quite different. They are quite high in Greece and very low in Belgium. The fifth country for which resistance data were obtained was France [19]. France had a unique aminoglycoside-usage pattern: equal (-30%) usage of gentamicin, netilmicin, and amikacin. This usage correlated with a high incidence of AAC(6' )-I (tobramycin, netilmicin, amikacin, dibekacin, and kanamycin), both alone and in combination with other aminoglycoside-modifying enzymes. Resistance rates in France were very similar to those observed in Italy except for resistance of P. mirabilis isolates, which was much lower in France. Thus, on the basis of this recent large survey [19], there does not seem to be any correlation between the aminoglycoside-usage pattern and rates of resistance in Enterobacteriaceae. Perhaps, as previously suggested [18], resistance rates are related to total aminoglycoside usage while aminoglycoside resis- Table 4. Comparison of aminoglycoside-usage groups in five European countries and the incidence of AAC(6')-I with aminoglycoside resistance rates from a 1995 survey [19]. Data from indicated country Source, variable West Germany Italy Belgium Greece France Present report Aminoglycoside-usage group (see figure 3) GT GTN TNA NA GNA No. of hospitals surveyed AAC(6')-I incidence (%) Alone Combined Survey [19] No. of hospitals surveyed Gentamicin resistance (%) Enterobacter cloacae Klebsiella pneumoniae Escherichia coli Proteus mirabilis

16 CID 1997; 24 (Suppl 1) Use of and Resistance to Aminoglycosides S61 tance mechanisms are related to the type of aminoglycoside used. Consistent with this is the general idea that total aminoglycoside consumption per capita in Greece, Italy, and France is higher than in Belgium or West Germany ([18] and personal communication, P. Mackey). However, these aminoglycoside resistance mechanisms are plasmid-mediated, and it seems likely that both the type and extent of use of other antibiotics must also play a role in the maintenance of resistance. In fact, the general correlation with total aminoglycoside consumption mentioned above may simply be a reflection of higher total antibiotic use in Greece, France, and Italy. Earlier studies [20] in which amikacin was the most common or only aminoglycoside used showed either no change in resistance rates or lower resistance rates. The finding during these studies that resistance rates did not change during a period of exclusive use of one aminoglycoside is consistent with the idea that these rates are related to either total aminoglycoside usage or total antibiotic consumption. The general idea that aminoglycoside resistance rates have remained similar since 1984 in most Western countries is also consistent with this hypothesis. Unlike the current studies, the studies in which only amikacin was used often failed to show an effect of aminoglycoside usage on aminoglycoside resistance mechanisms. These studies were carried out before 1984, however, usually after a period of amikacin usage of years. Only one of the studies carried out before 1984 and summarized in this article showed an effect of a change in aminoglycoside usage on aminoglycoside resistance mechanisms. This was the study in Japan [3], where amikacin was used much earlier than in the West. In addition, prior aminoglycoside usage was different than in the West, since it involved dibekacin and kanamycin much more than gentamicin. It seems likely that the changes in aminoglycoside resistance mechanisms after 1984 that are described in this article have resulted from a long-term change in aminoglycoside selective pressure (over 5 years or more). Changes in aminoglycoside selective pressure from exclusive gentamicin use began, in general, during with the introduction of tobramycin and amikacin and continued with the introduction in the early 1980s of netilmicin in Europe and Latin America. The earliest reports [4, 13] of changes in aminoglycoside resistance mechanisms occurred in Thus, it is not surprising that the introduction of a new aminoglycoside, such as occurred in Japan in 1988 when isepamicin was introduced and again in 1992 when arbekacin was introduced, may not as yet have affected either aminoglycoside resistance rates or aminoglycoside resistance mechanisms [21]. Study Coordinators and Individual Investigators with Large Collections Belgium, Luxembourg, Netherlands: Y. Glupczynski. Germany: A. Bauernfeind and S. Schweighart. England: K. Shannon. Poland: J. Patzer. Italy: G. Molinari and G. C. Schito. Spain: R. GOmez- Lus and S. GOmez-Lus. Portugal: H. Ferreira, J. C. Sousa, and M. J. M. Vaz. France: E. Collatz, R. Bismuth, T. Lambert, P. Courvalin, and C. Minozzi. South Africa: K. Klugman and Y. Bilgeri. Greece: H. Giamarellou and G. Petrikkos. Turkey: H. Akalin and D. Giir. Latin America: ACU; Argentina: M. Woloj, A. Rossi, J. Casellas, M. Tokumoto, and E. Couto. Chile: C. Juliet, M. E. Pinto, and R. Zemelman. Uruguay: W. Pedreira. Latin America: GMV; Guatemala: M. Fernandez. Mexico: I. Leal. Venezuela: M. Guzman, J. Murillo, P. Isturiz, and A. Merentes. Pan-Pacific: A. Bremner, B. Ho, K. Mayer, J. Ellal, W. Fu, and D. Zhu. Isepamicin clinical isolates: K. Dornbusch and E. Goransson. References 1. The Aminoglycoside Resistance Study Groups. The most frequently occurring aminoglycoside resistance mechanisms combined results of surveys in eight regions of the world. J Chemother 1995; 7(suppl 2): Miller GH, Sabatelli FJ, Naples L, Hare RS, Shaw KJ, the Aminoglycoside Resistance Study Groups. The changing nature of aminoglycoside resistance mechanisms and the role of isepamicin a new broad-spectrum aminoglycoside. J Chemother 1995; 7(suppl 2): Shimizu K, Kumada T, Hsieh W-C, et al. Comparison of aminoglycoside resistance patterns in Japan, Formosa, and Korea, Chile and the United States. Antimicrob Agents Chemother 1985; 28: Hare RS, Miller GH, Sabatelli FJ, Weiss WJ, Shlaes DM. Aminoglycoside (AG) usage and enzymatic mechanism of resistance [abstract 416]. In: Program and abstracts of the 25th Interscience Conference on Antimicrobial Agents and Chemotherapy. Washington, DC: American Society for Microbiology, 1985: Hare RS, Shaw KJ, Sabatelli FJ, et al. Survey of aminoglycoside resistance in 29 USA hospitals [abstract 675]. In: Program and abstracts of the 29th Interscience Conference on Antimicrobial Agents and Chemotherapy. Washington, DC: American Society for Microbiology, Dornbusch K, European Study Group on Antibiotic Resistance. In vitro susceptibility to aminoglycoside antibiotics in blood and urine isolates consecutively collected in twenty-nine European laboratories. Eur J Clin Microbiol 1987; 6: Dornbusch K, Miller GH, Hare RS, Shaw KJ, the ESGAR Study Group. Resistance to aminoglycoside antibiotics in gram-negative bacilli and staphylococci isolated from blood. Report from a European collaborative study. J Antimicrob Chemother 1990; 26: The Belgian Ag R Study Group, Glupczynski Y, Miller GH, Sabatelli FJ, Hare RS, Shaw KJ. Changes in aminoglycoside (Ag) resistance (R) mechanisms (M) in Belgium over 15 years [abstract C124]. In: Program and abstracts of the 36th Interscience Conference on Antimicrobial Agents and Chemotherapy. Washington, DC: American Society for Microbiology, Shimizu K, Nasu M, Tozuka K, Shaw K, Hare R, Miller G. Comparison of aminoglycoside (Ag) resistance patterns (AGRP's) in gram-negative bacteria (GNB) from the eastern (E) and western (W) areas of Japan [abstract 397]. In: Program and abstracts of the 27th Interscience Conference on Antimicrobial Agents and Chemotherapy. Washington, DC: American Society for Microbiology, 1987: Price KE, Kresel PA, Farchione LA, Siskin SB, Karpow SA. Epidemiological studies of aminoglycoside resistance in the USA. J Antimicrob Chemother 1981; 8(suppl A): Rather PN, Orosz E, Shaw KJ, Hare R, Miller G. Characterization and transcriptional regulation of the 2'-N-acetyltransferase gene from Providencia stuartii. J Bacteriol 1993; 175(20): Snelling AM, Hawkey PM, Heritage J, Downey P, Bennett PM, Holmes B. The use of a DNA probe and PCR to examine the distribution of