Prevalence and clonality of extended-spectrum β-lactamases in Asia
Corresponding author and reprint requests: P. M. Hawkey, Division of Immunity and Infection, University of Birmingham, Birmingham B15 2TT and Health Protection Agency, West Midlands Public Health Laboratory, Birmingham Heart of England NHS Trust, Bordesley Green East, Birmingham B9 5SS, UK
Asia is almost certainly a part of the world in which extended-spectrum β-lactamases (ESBLs) have emerged de novo, with some early antimicrobial resistance studies showing high levels of the ESBL phenotype, particularly among Klebsiella, and most notably in China, Korea, Japan and India. There is a lack of genotyping studies but work from the late 1990s suggests that SHV-5 and SHV-12 were most common then, with only very rare reports of TEM-related ESBL genes. As in other parts of the world, quite marked differences have since been seen in the pattern of ESBL genes, particularly in relation to the CTX-M family. The early emergence of TOHO CTX-M-2 in Japan contrasted with CTX-M-3 and -14 in China and many other parts of the Far East, suggesting the separate transfer of genes from the genome of Kluyvera spp. to mobile genetic elements in human-associated Enterobacteriaceae. ESBL production rates are now very high compared with Europe. In most countries, there are mixtures of CTX-M types, with VEB appearing significantly in Vietnam and Thailand, and ESBL isolates from India being completely dominated by the presence of blaCTX-M-15 alone, with no other CTX-M types reported. With the total population of India and China being c. 2.4 billion and with faecal carriage rates of, probably, c. 10%, these countries represent major reservoirs of blaCTX-M genes. Increasing international travel and trade will lead to the movement of many of these ESBL genes. The high prevalence of ESBL genes in Asia means that the empirical treatment of serious infections with β-lactam antibiotics, except carbapenems, is seriously compromised.
Asia probably has a long history of the occurrence of extended-spectrum β-lactamase (ESBL)-producing bacteria. Excepting a few centres, interest in ESBLs in the region is a comparatively recent phenomenon and, as in other regions of the world, coverage has not been comprehensive, particularly in relation to genotypes. It is well-recognised that there can be marked variation in the incidence and or genotype of ESBLs in hospitals close to one another, and certainly among countries .
There were no comprehensive reports on the incidence of ESBLs from countries in the region in the 1980s and early 1990s. There were, however, a number of sporadic reports of ESBLs, notably of the SHV-2 type, from China in 1988 , and the Toho-1 ESBL produced by Escherichia coli from Japan in 1993 . The recognition of the importance of ESBLs as a major mechanism of β-lactam resistance throughout the region came with presentation of data from the 1998–1999 SENTRY antimicrobial surveillance programme . The incidence of ESBL production (no genotyping was undertaken) among E. coli isolates in the four Chinese sites varied from 13% to 35%. Rates ≥20% for the ESBL phenotype in Klebsiella pneumoniae in all participating mainland Chinese centres (one reaching 60%), in one each of three Japanese and Taiwanese centres, and in the single Singapore centre and Philippines centre, were confirmed. Such high rates had previously been reported only from South America; in a follow-up study (1998–2002), lower rates were found in K. pneumoniae isolates from Australia and Japan (≤10%), but that in China was 30% . The other area of Asia with high reported rates of ESBL production is the Indian subcontinent. One study reported a 68% prevalence of ESBL phenotypes among E. coli and K. pneumoniae isolates, one of the highest rates reported for any country worldwide . A number of other studies have reported high rates in India , with a similar situation existing in Pakistan .
Although SHV-type ESBLs apparently dominated most studies in Asia, where genotyping was first undertaken in the 1990s , the data are sparse because so few isolates from that era were examined. It is certainly true that SHV-5 and -12 enzymes were reported in early studies from Japan and Korea and remain quite common in more recent studies [10–12]. But as in Europe, it is the emerging dominance of CTX-M β-lactamases that has changed the nature and significance of ESBL production in Enterobacteriaceae. It seems likely that, particularly in China and India, where high rates of ESBL production were reported in the 1990s, CTX-M enzyme production had expanded from the very early 1990s or late 1980s. Interesting regional and national variations apparently exist in blaCTX-M genotypes, although a number of studies only report ‘group’ genotyping. In China, and a number of other Asian countries, the dominant types are CTX-M-14 and CTX-M-3, but Japan had a different profile, with CTX-M-2 being a common type. . These epidemiological aspects are considered below.
Although one of the early reports of SHV-2 enzymes was from China , there were very few further reports of ESBL genotypes until the first description of CTX-M-13 and CTX-M-14 (one of the most numerous genotypes in Asia) from isolates of four different species collected from Guangzhou in 1997–1998. In total, 12/15 isolates with ESBLs from this study carried blaCTX-M genes, CTX-M-14 being the most common, followed by CTX-M-9, first identified in 2000 from Spain . In that study too, SHV-12 was reported for the first time in China, while a subsequent large-scale study of isolates from six provinces, collected between 1998 and 2002, reported SHV-12 as the most common SHV type (ten isolates), with much smaller numbers of SHV-5, -2 and -9 . Nevertheless, the 17 isolates from that study with blaSHV were dwarfed by the 288 isolates of E. coli and 142 isolates of K. pneumoniae that produced CTX-M β-lactamases. In both species, CTX-M-14 (271 isolates) dominated, closely followed by CTX-M-3 (70 isolates), a common finding in other Chinese centres, as reviewed by Munday et al. . Yu et al.  reported some rarer genotypes of blaCTX-M, namely CTX-M-13, -22, -27 and -28, mainly as single isolates; the third most common genotype was CTX-M-24 (35 isolates), which is closely related to CTX-M-14.
High rates of ESBL production (with no genotyping) have been reported from various hospital centres in China. Rarer types of ESBL have not been reported, except for the first description of VEB-3 β-lactamase in 12/27 isolates of Enterobacter cloacae from Shanghai (nine isolates also carried a blaCTX-M-3-like gene); otherwise, VEB ESBLs are most common in Vietnam and Thailand . ESBLs were reported very early from Hong Kong, with rates of 1.6% in E. coli and 2.6% in K. pneumoniae in 1990 . Subsequently, rates rose rapidly in mainland China, whereas they have remained lower in Hong Kong, at 11% in E. coli and 13% in K. pneumoniae in 1999 .
The occurrence of ESBLs in Taiwan has been well-documented and is reviewed in a recent publication . Relatively high incidence figures (for that time) for ESBL phenotypes were reported for three centres in northern Taiwan, contributing to the 1998–2002 SENTRY programme, with overall rates of ESBL production of 13.5% in K. pneumoniae and 5.6% in E. coli . The increase over time in ESBL-producing K. pneumoniae isolates is illustrated in a single centre in northern Taiwan, where they increased from 3.4% in 1993 to 10.3% in 1997 . The distribution of ESBL genotypes in Taiwan is very similar to that of mainland China, with early reports of SHV-5/12 and β-lactamases, followed by a growing dominance of isolates with CTX-M-3/14 β-lactamases . As in most other Asian countries, in Taiwan TEM ESBLs are rare, with only TEM-10 having been reported .
The first report of ESBL-producing bacteria from Taiwan was prompted by a marked increase in ceftazidime resistance among K. pneumoniae (to 30% of all isolates) in 1997 . Isoelectric focusing of β-lactamases was used; pIs above 7.0 were found in all cases, and it was presumed that ESBLs were SHV types. DNA sequencing of PCR products from SHV-specific primers revealed 22 isolates with SHV-5 and two with SHV-2; seven others had pIs of either 7.9 or 7.75, and it is likely that these were unrecognised CTX-M types . Subsequent surveys have demonstrated variations in the distribution of genotypes among species of Enterobacteriaceae, with CTX-M-3/SHV-12 in K. pneumoniae and CTX-M-3 in E. coli, Serratia marcescens and non-typhoidal Salmonella . CTX-M-14 is less frequently seen than in mainland China but is common in Proteus mirabilis , and SHV-12 is the most common ESBL in E. cloacae .
The evolution of different genotypes of SHV ESBLs has been postulated to be due to the accumulation of mutations in either SHV-1 or -11 (both non-ESBLs) . The relative abundance of SHV ESBLs in Taiwan results from their early emergence and mass spread of the common successful genotypes (e.g., SHV-5/12). One study supports the notion that SHV-2/-5/-12, and rarer types, SHV-26 and the non-ESBL type SHV-25, have evolved by stepwise mutation, the latter two possibly from SHV-1 and SHV-11, respectively .
One of the earliest documented reports of the occurrence of ESBL phenotypes comes from Korea, with 7.5% of E. coli and 22.8% of K. pneumoniae isolates at Yonsei Medical Center being described as ESBL-positive in 1994 . Interesting subsequent studies on genotypes in the late 1990s showed a pattern very different from that of other Asian countries. Unusually, TEM-52 was one of the most frequently encountered ESBLs in E. coli and K. pneumoniae from three centres in 1996, produced by 22/53 isolates, often in combination with other genotypes, against 26/53 with SHV-12 and 14/53 with SHV-2a . When 36 blood culture isolates of E. coli and K. pneumoniae from 1994–1996 were characterised, the majority produced TEM-52 but two produced TEM-15 and two produced TEM-88. No particular hydrolytic kinetic advantages were found for TEM-52 and TEM-88 over the parental type TEM-15 .
A retrospective study of isolates from Seoul, Korea in 1995–1996 revealed one isolate of E. coli and two of K. pneumoniae producing CTX-M-14 , an enzyme first described in Guangzhou, China in 1997–1998 . The prevalence of bacteria with CTX-M ESBLs in Korea has since grown but not to the extent seen in mainland China. A survey in 2005 reported 23/246 E. coli isolates and 55/239 K. pneumoniae isolates with an ESBL phenotype, and overall, only 26 had CTX-M enzymes . Strains with CTX-M-14 were not common (13 reported), and those with CTX-M-3 and CTX-M-15 were even less common. It should be noted, in this context, that CTX-M-15 differs from CTX-M-3 by only one amino acid substitution, Gly240Asp, which could be a de-novo evolutionary event. However, as CTX-M-15 has probably been present in India for some considerable time, and is present in both E. coli and Klebsiella spp. at a high frequency, spread from India is more likely . Although blaCTX-M-15 has subsequently spread to many parts of the world, it remains very rare in Asia, supporting this hypothesis . CTX-M-9 was also reported in isolates from three Korean hospitals in 2003 , and recently, CTX-M-12, which has three amino acid changes from CTX-M-3, was reported. It has been reported previously (and rarely) only from Kenya, Colombia and China . GES-3, a rare ESBL previously reported only from Greece and Japan, has also been found in Korea .
As in Korea, the pattern of ESBL genotypes in Japan is quite different from that seen in surrounding countries, although universally successful types, e.g., CTX-M-14, have recently become more common. Although some surveys suggest that the overall rate of ESBL phenotypes is comparatively low (6.5% in E. coli, 11.0% in K. pneumoniae ) ESBLs have been present for a long time. Studies by Kawakami et al.  from a hospital in Tokyo reported rates of 0.4% and 0.6% in E. coli and K. pneumoniae, respectively, in 1990, rising to 1.7% and 7.2% in 1995. A survey of seven laboratories in the Kinki region, over 2-month periods in 1998 and 2000, revealed 10% of E. coli isolates and 4.0% of K. pneumoniae isolates to have an ESBL phenotype. Molecular genotyping revealed 13 isolates of S. marcescens producing CTX-M-3 and CTX-M-3, and SHV-12 was found in K. pneumoniae and E. coli . In a survey of 426 urinary isolates of E. coli from patients in 37 hospitals on Northern Kyushu Island between January and September 2003, the overall ESBL rate was 14%. Sixteen hospitals reported ESBLs, and among these, ≥40% of isolates produced ESBLs . The diversity of rates probably reflects differences in cross-infection rates and in local antibiotic usage. The most recent countrywide survey, unfortunately, does not include denominator data .
Two very early members of the CTX-M group were identified in Japan: TOHO-1 and -2. These have not spread or evolved de novo outside Japan, except for one reported single isolate producing TOHO-1 from Argentina . TOHO-1 (CTX-M-44) is very closely related to CTX-M-2, and was reported in relation to an isolate of E. coli collected in 1993 [1,3]. TOHO-2 (CTX-M-45) was described a short time later and is only distantly related (by amino acid sequence) to CTX-M-9. TOHO-1 (CTX-M-44) and TOHO-2 (CTX-M-45) have recently been given numbers in the CTX-M series (http://www.Lahey.org-internet-studes-inc_webt.asp). The precise frequency of occurrence of TOHO-1 (TOHO-2) has never been reported in epidemiological studies, and the veracity of the sequence has been questioned , as it is so closely related to CTX-M-2, and many studies in Japan have used group-specific PCRs or DNA probes that would not differentiate among these genotypes. A major survey from Tokyo in 1997–1998, undertaken to ascertain the presence of SHV- and TEM-derived ESBLs, revealed 28/16 805 E. coli isolates and 41/9794 K. pneumoniae isolates with ESBL phenotypes . TOHO group enzymes were more common in E. coli, and SHV-12 enzymes, in particular, were more common in K. pneumoniae, while only a single example of a TEM ESBL, TEM-26, was found. A much more recent survey of the whole of Japan again used only group-specific PCR to identify CTX-M genotypes, but again confirmed the frequent occurrence of the CTX-M-2 group. The CTX-M-9 group (possibly CTX-M-14) and the CTX-M-1 group were almost as frequent as the CTX-M-2 group . All isolates of P. mirabilis and some Acinetobacter spp. produced only CTX-M-2, possibly due to the limited range of the plasmid host . In this study, genotyping using PCR for blaCTX-M was carried out only on 317 isolates, phenotypically thought to carry blaCTX-M (more resistant to cefotaxime than ceftazidime). In total, 1456 Gram-negative bacilli that were resistant to oxyimino-cephalosporins, from 132 hospitals, were submitted for study; those showing other patterns of resistance or a negative double-disk-diffusion test were excluded from study, as potentially leading to bias  The spread of CTX-M-2 into Acinetobacter is a worrying development and may relate to its carriage by a class 1 integron, facilitating transfer among different replicons. A report of cross-infection in a neurosurgical unit among three patients, due to Acinetobacter baumannii isolates carrying a plasmid encoding CTX-M-2, all of which had a restriction endonuclease pattern identical to a plasmid in a contemporaneous P. mirabilis isolate, demonstrates the importance of plasmid transfer in the spread of blaCTX-M .
India and pakistan
Both India and Pakistan have reported high rates of ESBLs since the 1990s [6,8]. ESBL-producing Enterobacteriaceae are also well-established in the community, as evidenced by a study of patients with infections and no prior hospitalisation in the preceding 3 months. Three hundred isolates were collected, 72 (24%) being ESBL producers, 50 being from patients in the community; significantly, 74% and 76% had a history of prior use, at some time, of a cephalosporin and quinolone, respectively . There have been relatively few genotyping studies, but SHV-12 in K. pneumoniae has been reported from Southern India to be linked to the qnrB plasmid-mediated quinolone resistance determinant, the first report being from S. marcescens [41,42] There is a report of SHV-5 produced by Salmonella senftenberg causing an outbreak on a burns ward in Delhi . The only genotype of a TEM ESBL is TEM-104, reported in ten isolates of K. pneumoniae from New Delhi in 2001–2002 . CTX-M-15 β-lactamase is now widely distributed across the world, but was first described in a small number of isolates from Delhi in 2000 . A very recent survey from three widely dispersed centres in India showed that 95/130 cefpodoxime-resistant E. coli and K. pneumoniae isolates obtained between 2003 and 2005 carried a blaCTX-M gene, which, when genotyped using reverse-line hybridisation  and DNA sequencing, was shown to be blaCTX-M-15 in all cases . The study by Ensor et al. prompted a letter reporting on isolates of Klebsiella spp. and E. coli collected in the late 1990s from six widely dispersed centres; 47 isolates were examined using PCR and DNA sequencing, and 37 were found to carry blaCTX-M-15, which was the only CTX-M genotype found . CTX-M-15 was also identified from two of the K. pneumoniae isolates collected during 2002–2003 in Coimbatore in South India . India, therefore, not only appears to have very high rates of ESBL production across the country, but is entirely dominated by blaCTX-M-15, with no other blaCTX-M genotypes having yet been detected.
Surveys of ESBLs from Pakistan have, to date, only reported on the occurrence of ESBL phenotypes. An early survey by Zaman et al., of 200 nosocomial isolates of Enterobacteriaceae, reported a rate of 35% . A comprehensive study of both hospital (Aga Khan Hospital, Karachi, Pakistan) and community isolates of E. coli, Klebsiella spp. and Enterobacter spp. revealed ESBL phenotype rates of 41%, 36% and 50%, respectively; the overall rate for the 2840 isolates collected between April and October 2002 was 40% . The rate of isolation of ESBLs was highest among inpatients (52%) and lowest in patients from community health centres (30%). At the time of writing, there have been no genotyping studies. A recent paper suggested the identification of CTX-M-producing bacteria, but this was based on the fact that the MIC of cefotaxime is higher than that of ceftazidime in four isolates of E. coli, and no molecular data were reported .
Other asian countries
High rates of the ESBL phenotype (26% of Enterobacteriaceae) were reported as early as 1994–1996 in Khon Kaen, northern Thailand . A genotyping study of those isolates found that 26/43 produced SHV-12, 13/43 SHV-5, 2/43 SHV-2a and 1/43 VEB-1 . VEB-1 is always reported to be encoded on a cassette as part of a class I integron; in addition to being reported from several species of Enterobacteriaceae, blaVEB-1 has been found in multiple isolates of Pseudomonas aeruginosa in Thailand, illustrating its considerable genetic mobility . The VEB series of β-lactamases was originally described in Vietnam (hence, Vietnamese extended-spectrum β-lactamase), and involved 14 of 55 ESBL-producing Enterobacteriaceae. While 14 carried blaCTX-M-like genes, DNA sequencing of Klebsiella spp. isolates revealed six of these genes to be blaCTX-M14 and two its close relative, blaCTX-M17, with blaSHV-2 being found in 21 isolates . There are few detailed reports of ESBLs from other Asian countries, but a recent study from Malaysia reported blaSHV-5 to be the dominant ESBL there .
Asia, like many other regions of the world, has experienced a significant increase in the numbers of ESBL-producing bacteria in the new millennium. The current rates in some countries are very high and there are clues that the CTX-M enzymes, which are responsible for much of the increase, have been present since the late 1980s. A recent study of isolates of Enterobacteriaceae causing community-acquired infection in multiple Chinese centres revealed surprisingly high rates of ESBL production (E. coli 16%, Klebsiella spp. 17%) . This, combined with disquieting evidence that food animals studied in Hong Kong (but largely raised on the mainland) carried the same distribution of CTX-M genotypes as is seen in human infections , suggests that these resistance genes are set to become as common as blaTEM-1. Faecal carriage rates of ESBL producers are likely to be high, and the spread of Enterobacteriaceae through poor-quality drinking water and poor sewage disposal probably accounts for this situation. With the populations of India and China standing at 1.1 billion and 1.4 billion, respectively, these two countries surely represent the largest reservoirs of CTX-M ESBL genes in the world. Increasing travel and trade will contribute to the worldwide spread of locally evolved CTX-M genotypes.
I thank my colleagues and students, present and former, for their hard work and illuminating discussions during our studies of ESBL-producing bacteria from Asia. Support from the British Society for Antimicrobial Chemotherapy, British Council, University Grant Commission, India, Royal Thai Government, Merck and Wyeth for work undertaken by Professor Peter Hawkey is acknowledged.