Guidelines for the validation and application of typing methods for use in bacterial epidemiology
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Guidelines for the validation and application of typing methods for use in
bacterial epidemiology
A. van Belkum1, P. T. Tassios2, L. Dijkshoorn3, S. Haeggman4, B. Cookson5, N. K. Fry6, V. Fussing7,
J. Green8, E. Feil9, P. Gerner-Smidt10, S. Brisse11 and M. Struelens12 for the European Society of Clinical
Microbiology and Infectious Diseases (ESCMID) Study Group on Epidemiological Markers (ESGEM)
1
Erasmus MC, Department of Medical Microbiology and Infectious Diseases, Rotterdam, The
Netherlands, 2National and Kapodistrian University of Athens, Department of Microbiology, Athens,
Greece, 3Leiden University Medical Center, Department of Infectious Diseases, Leiden, The Nether-
lands, 4Swedish Institute for Infectious Disease Control, Department of Bacteriology, Solna, Sweden,
5
Laboratory of Health Care Associated Infections, 6Respiratory and Systemic Infection Laboratory,
Health Protection Agency, Centre for Infections, London, UK, 7Novo Nordisk, QC Microbiology SDK,
Novo Alle, Bagsvaerd, Denmark, 8Statistics, Modelling and Bioinformatics Department, Health
Protection Agency, Centre for Infections,London, 9University of Bath, Department of Biology, Bath,
UK, 10Centers for Disease Control and Prevention, Foodborne and Diarrheal Diseases Branch, Divison
of Bacterial and Mycotic Diseases, Atlanta, GA, USA, 11Institut Pasteur, Unit BBPE28, Paris, France and
12
Université Libre de Bruxelles, Hôpital Erasme, Bacteriologie,Brussels, Belgium
ABSTRACT
For bacterial typing to be useful, the development, validation and appropriate application of typing
methods must follow unified criteria. Over a decade ago, ESGEM, the ESCMID (Europen Society for
Clinical Microbiology and Infectious Diseases) Study Group on Epidemiological Markers, produced
guidelines for optimal use and quality assessment of the then most frequently used typing procedures.
We present here an update of these guidelines, taking into account the spectacular increase in the
number and quality of typing methods made available over the past decade. Newer and older,
phenotypic and genotypic methods for typing of all clinically relevant bacterial species are described
according to their principles, advantages and disadvantages. Criteria for their evaluation and
application and the interpretation of their results are proposed. Finally, the issues of reporting,
standardisation, quality assessment and international networks are discussed. It must be emphasised
that typing results can never stand alone and need to be interpreted in the context of all available
epidemiological, clinical and demographical data relating to the infectious disease under investigation.
A strategic effort on the part of all workers in the field is thus mandatory to combat emerging infectious
diseases, as is financial support from national and international granting bodies and health authorities.
INTRODUCTION
CENTRAL THEME
Bacterial typing methods generate isolate- The ability to quickly and reliably differentiate
specific molecular fingerprints for among related bacterial isolates is essential for
assessment of epidemiological relatedness epidemiological surveillance, and is an endeav-
our as old as the discipline of bacteriology itself.
Long-standing ‘conventional’ typing methods,
such as bacteriophage typing of Staphylococcus
aureus and Listeria monocytogenes [1,2], serotyping
of Salmonella spp. and Escherichia coli [3,4], or
biochemical typing of Enterobacteriaceae [5],
have historically been important contributors to
Corresponding author and reprint requests: A. van Belkum,
Erasmus MC, Department of Medical Microbiology and Infec-
our understanding of the natural history and
tious Diseases, Dr. Molewaterplein 40, 3015 GD Rotterdam, The epidemiology of infections caused by strains
Netherlands. E-mail: a.vanbelkum@erasmusmc.nl of these clinically relevant bacterial species.
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–46
2
Similarly, antibiogram typing has for many years evolutionary history. For example, two isolates
been and, as a matter of fact, still is, in the field of that are identical according to phage typing
clinical microbiology, a first-line method to iden- might in fact be quite unrelated, and conversely,
tify possible cases of bacterial cross-transmission two isolates that show quite different phenotypes
in healthcare institutions. These methods for for a single marker might in fact be closely
bacterial phenotyping have a clear purpose in related. For these reasons, phenotyping has been
the confirmation and elucidation of local and largely replaced by genotypic or ‘molecular’
national healthcare-associated outbreaks due to typing over the past two decades [8–13]. In
bacterial strains [1]. However, although still principle, at least, asexual (clonal) reproduction
useful for specific purposes, they have a number by binary fission implies that genotypic markers
of practical limitations which render them unsuit- should reflect evolutionary history and would
able for comprehensive studies of bacterial popu- therefore be useful in delineating a natural
lation structure and dynamics, and also for the taxonomy. In practice, the ease with which genes
scientifically less ambitious, but very critical, can be transferred among different lineages
endeavours of infection control and surveillance means that the data from multiple markers are
[6,7]. Furthermore, most phenotypic methods required, and even then there is no guarantee
have been developed for specific bacterial species that a natural taxonomy will present itself
and are not generally applicable. However, [14]. Polyphasic taxonomy currently uses com-
although it is generally accepted that phenotyp- binations of different phenotypic or genotypic
ing cannot usually stand alone, in some cases datasets to define genera, species and even
(e.g., serotyping of salmonellae), it is a very taxonomically relevant subspecies [15–18]. At
useful prerequisite. Nevertheless, the develop- the same time, however, there are inherently
ment, application and quality control of phage polymorphic loci present in the genomes of all
typing and serotyping are labour-intensive and bacterial species that enable further subspecies
require skills and methodologies that are difficult differentiation. Thus, DNA typing, which essen-
to maintain at levels of quality sufficient to satisfy tially comprises the direct or indirect assessment of
the standards of today’s accreditation bodies for subspecies nucleotide sequence motifs and their
microbiology laboratories. More importantly, any variation in both primary structure and number of
given phenotype does not always accurately copies per chromosome (see Fig. 1 for a generalised
reflect the genotype of a microorganism, and scheme), can reproducibly reveal conserved as well
therefore may not provide a reliable and stable as variable characteristics, both at different taxo-
epidemiological marker. The rate of genetic nomic levels and at levels below species/subspe-
exchange within many bacterial species means cies, the lowest taxonomic rank with official
that a given phenotype may not always reflect standing in nomenclature.
Figure 1. The general features of molecular typing methods. The four boxes show the various molecular concepts
associated with genetic variability. Below these boxes, the typing techniques most suited for the detection of such nucleic
acid changes are indicated. More technical detail can be retrieved from various sections in the text.
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–46
3
Unfortunately, new molecular typing methods public health, and workers in reference labora-
are often proposed for general use without suffi- tories.
cient prior critical evaluation. For example, they
may not have been standardised, a minimal
DEFINITIONS REGARDING ISOLATE
number of isolates may have been used for
RELATIONSHIPS
validation, their agreement with epidemiological
data may not have been assessed, or the suitabil- Bacterial typing has acquired its own vocabulary,
ity of a specific method–microbe combination for in part borrowed from that of other scientific
a specific bacterial taxon may not have been disciplines, including population biology, molec-
addressed [19–28]. Finally, basic terminol- ular biology, taxonomy and ecology. Use of this
ogy—including fundamental terms such as ‘iso- terminology is not always consistent and can be
late’, ‘strain’, ‘type’ or ‘clone’—is often used confusing. Prior to presentation of a glossary, we
differently by different workers in the field of would like to discuss the terms ‘isolate’, ‘strain’
bacterial epidemiology. and ‘clone’ in detail, in order to highlight some of
Here, we present an update of the previous the debatable issues concerning definitions, and
ESGEM guidelines for the correct application thereby suggest a more standardised and uniform
of methods and interpretation of the resulting terminology.
data [29]. We endeavour to define the terminol- The terms ‘isolate’ and ‘strain’ are often used
ogy used in microbial typing, distinguish the interchangeably, but not always appropriately. A
major means and purposes of bacterial typing, bacterial isolate can be defined simply as a single
provide criteria for evaluation, and outline the isolation in pure culture from a clinical speci-
advantages, limitations and unresolved issues men. Depending on the state of characterisation,
related to the methods currently used. We an isolate may be referred to as, for example,
intend to increase awareness of the importance ‘urine isolate X’ (if only the sample type is
of methodological evaluations and optimisa- known) or ‘MRSA isolate Y’ (if the species and
tions, and the appropriate use of control and some antimicrobial resistance properties are
reference strains, as well as prudent data inter- known). Ultimately, isolates can be characterised
pretation. In short, we aim to define the pur- as descendants of the same strain. However,
pose and choice of methods, in combination there is no agreement concerning the minimal
with interpretation of the results, thereby facil- sets of characters required to define any kind of
itating the development of practical decision strain. A reference strain is a well-characterised
trees. We suggest useful ways for the commu- strain that is maintained in pure culture for
nication of typing data in general, and more further study, while a type strain is a special
specifically, communication from the laboratory kind of reference strain, i.e., the strain with
to the clinic. We include discussions on differ- which the name of the species is permanently
ent typing applications and their globalisation, associated. An isolate can be assigned to a
and, importantly, on quality control. Finally, the defined type according to the results of the
links between practical baterial typing and application of a particular typing method, e.g.,
phylogeny, population biology and taxonomy pulsed-field gel electrophoresis (PFGE) type X,
are considered. This position paper has been spa type Y. It must be noted that isolates with
developed through interactions with microbiol- identical typing results need not necessarily
ogists active in the field, and aims to propose belong to the same strain, since different strains
genuine and applicable general typing guide- may be indistinguishable with respect to a
lines. These guidelines, however, should always typing method. The opposite can also be true;
be applied carefully and their consequences isolates with different types may be part of the
interpreted critically in all instances. The same (pandemic) strain. This can be observed
intended audience includes, among many others, when the intrinsic evolutionary clockspeed of a
general and clinical microbiologists, infectious given species is higher than average. At present,
disease specialists, infection control managers, different nomenclatures for bacterial strains,
higher degree students, research technologists isolates and types exist and these must be
interested in the molecular epidemiology of considered with care and used appropriately.
bacteria, decision-makers in the context of To ensure the consistent use of the terms ‘isolate’
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–46
4
and ‘strain’, we suggest the following example:
two isolates (1 and 2) can be representatives of GLOSSARY OF TERMS
one strain (A), but two strains (A and B) can
never be the same isolate (1). Some of the general terms defined below have
The terms ‘strain’ and ‘clone’ are also used been previously described in the literature
interchangeably. The ‘clone’ concept, which is [29,43,44]. The internet was scanned via the
frequently used in the context of bacterial epi- Google search engine, using the terms as key
demiology and population genetics, also illus- words (search period November 2006). These
trates the importance of correct usage of definitions may have been adapted slightly to
definitions and nomenclature. ‘Clone’ is a term make them consistent with technological and
coined in the early 20th century in the field of philosophical approaches.
botany and used to denote a group of isolates Alert organisms: Bacterial species, strains,
descended from a common ancestor as part of a types or clones of special epidemiological
usually direct chain of replication [30,31]. The significance because of their predictable trans-
clonal relatedness of isolates is manifested by missibility and potential for causing difficult-
their display of a significantly higher level of to-treat infections. Identification of such an
similarity in their genotype and/or phenotype organism should alert healthcare providers
than can be expected for randomly occurring and trigger additional control measures such
and epidemiologically unrelated isolates of the as barrier isolation of colonised or infected
same species. This epidemiological working def- patients. Alert organisms are usually impor-
inition is less stringent than the definitions of a tant nosocomial pathogens or organisms with
clone used by microbial geneticists [31–35]. The an unusual antibiotic susceptibility profile.
interest in clones has increased over the past Bacterial epidemiology: The study of the
decades, due to the emergence of multiresistant dissemination of human bacterial pathogens,
or highly virulent clones of pathogenic bacteria including their transmission patterns, risk-
that have become widespread and seem to factors for and control of infectious disease in
remain stable for prolonged periods [24–26,33– human populations.
38]. Ørskov and Ørskov [31] proposed the Clonal complex: A group of bacterial isolates
following formulation: ‘The word clone will be showing a high degree of similarity, ideally
used to denote bacterial cultures isolated inde- based on near-identity of multilocus enzyme
pendently from different sources, in different profiles and multilocus sequence types. Clonal
locations, and perhaps at different times, but still complexes are identical to clonal groups.
showing so many identical phenotypic and Clonal reproduction: Mode of, usually, asex-
genotypic traits that the most likely explanation ual reproduction in which the offspring are
of this identity is a common origin.’ The opposite essentially identical to the parent. In bacteria,
of clonality is called panmixis, reflecting free clonal reproduction proceeds by binary fission.
DNA recombination among isolates [35,39,40]. Clone: Bacterial isolates that, although they
Examples of panmictic bacterial species are may have been cultured independently from
Helicobacter pylori [41] and Neisseria meningitidis different sources in different locations and
[42]. Isolates of panmictic bacterial species tend perhaps at different times, still have so many
to display extensive genetic variability, and the identical phenotypic and genotypic traits that
molecular fingerprints of a single strain may the most likely explanation for this identity is a
vary within a limited number of generations. common origin within a relevant time span.
Since the terms ‘isolate’, ‘strain’, ‘type’ and Cluster analysis: Comparative analysis of
‘clone’ have not always been used according to typing data collected for a variety of bacterial
the definitions given above, we propose defini- isolates in order to group the organisms
tions of a range of terms that are often used by according to their similarity in these data.
bacterial typists. We hope that these definitions Clusters can be identified by manual (visual)
will contribute to consistent usage among or computerised methods. The partitioning of
typists and scientists from affiliated fields such a dataset into subsets (clusters) reveals groups
as taxonomy and population genetics and that share common traits.
dynamics.
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–465
Comparative typing: A typing strategy aim- Fingerprint: A specific pattern (e.g., DNA
ed at assessing relatedness within a set of banding pattern) or set of marker scores (e.g.,
isolates without reference to other isolates. absorbance values) displayed by an isolate on
Convergence: Independent evolution along application of one or more typing methods.
parallel paths in unrelated lineages that ren- These fingerprints may be used for assessment
ders the lineages similar for some trait. of epidemiological relatedness among bacterial
Definitive (library) typing: Type allocation isolates.
of organisms according to an existing typing Fitness: The performance of a bacterial
scheme aimed at the development of isolate/strain in a particular environment in
(exchangeable) databases for long-term retro- terms of survival and reproductive rates.
spective and prospective multicentre studies Genetic drift: The process of random sam-
as well as epidemiological surveillance studies. pling of alleles for each generation, which is
Dendrogram: Binary tree illustrating a clus- relatively important in small populations, and
ter analysis performed on a number of isolates is an alternative evolutionary force for natural
for any chosen number of typing data. Each selection, causing allele frequencies to change.
tree, depending on the cluster algorithm used, Genetic drift determines the distribution of
depicts possible relationships between the alleles in different generations.
isolates included in the analysis. The basis for Genome: The complete genetic information
the tree is all the pairwise comparisons among of an organism as encoded in its DNA and/or
the included isolates. RNA.
Endemicity: Constant presence in a com- Genotype: Genetic constitution of an organ-
munity at a significant frequency, typically ism as assessed by a molecular method.
restricted to, or peculiar to, a locality or region. Hierarchical clustering: A method that
This usually presents as persistent occurrence emphasises how adjacent spatial units with
of disease in a population with a stable long- high or low disease rates might cluster by
term pattern of incidence around short-term ranking the units by disease rate, and then
stochastic fluctuations. examining how probable cluster adjacencies
Endemic: Strain present in a given setting would be compared to random conditions, and
over a longer period than if it were epidemic, marking off successive clusters wherever low-
although possibly at a relatively low frequency. probability values occur.
Epidemic: The occurrence of an organism Isolate: A population of bacterial cells in
above the usual endemic level as evidenced by a pure culture derived from a single colony. In
larger than expected number of infections. Used clinical microbiology, isolates are usually
as an adjective, the rapid and extensive spread derived from the primary culture of a clinical
by infection and/or colonisation that are widely specimen obtained from an individual patient.
prevalent, i.e., affecting many individuals in an Lineage: Group of isolates sharing essential
area or a population at the same time. characteristics due to common descent.
Epidemic strain: A strain that is suddenly Linkage disequilibrium: Non-random re-
present in a given setting with an unexpect- assortment of alleles occurring at different loci
edly high incidence. (However, it is sometimes due to physical linkage, usually due to lack or
difficult to determine whether increased inci- inhibition of recombination; strong in clonal
dence is due to strain traits, since there may organisms and absent in freely recombining
well be other explanations, e.g., poor hygienic populations.
conditions.) Mutation: The simplest mutation (change) in a
Evolutionary or phylogenetic tree: A dia- DNA or RNA sequence is a point mutation (a
gram that depicts the hypothetical phylogeny one-nucleotide change); other mutations include
(evolutionary history) of the taxa under con- deletion or insertion of one or more nucleotides.
sideration. The points at which lineages split Niche: A unique environment or set of
represent ancestor taxa to the descendant taxa ecological conditions in which a specific
appearing at the terminal points of the tree. (micro)organism occurs and thrives.
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–466
Outbreak: Local, initially small-scale, cluster Strain: The descendants of a single isolation
of disease generally caused by increased fre- in pure culture, usually derived from a single
quency of infection in a distinct population initial colony on a solid growth medium.
(may be caused by single epidemic strains or A strain may be considered an isolate or group
combinations of different strains). of isolates that can be distinguished from other
Panmixis: Situation in which gene exchange isolates of the same genus and species by
occurs randomly in the population at a high phenotypic and genotypic characteristics. Cul-
rate. Isolates of panmictic bacterial species tures of a particular microorganism, isolated at
(e.g., H. pylori and N. gonorrhoeae) tend to the same time from multiple body sites of a
display extensive genetic variability, and abso- patient and indistinguishable by typing, also
lute fingerprint identity may vary even within represent a single strain.
limited numbers of generations. Taxonomy: Theoretical study of organism
Pathogenicity: Biological ability to cause classification, which involves the sequential,
disease. interrelated activities of allocation of organ-
Pattern analysis: The process of comparing isms to taxa, their nomenclature and identifi-
data patterns generated by one or more typing cation.
methods. Type: A bacterial isolate may be allocated to
Phenotype: The observable characteristics of a named type according to an existing typing
a bacterial isolate/strain. Primary phenotype scheme. Type designations aim at facilitating
markers are the distribution of proteins and the handling and communication of typing
other cell components and the morphology results, and the development of (exchangeable)
and behaviour of cells. databases for long-term retrospective and pro-
Phylogeny: Evolutionary relationships spective multicentre studies, as well as epide-
among members of the same taxon (species, miological surveillance studies.
strains, etc.). Type strain: A strain, maintained in pure
Population: A group of organisms of the culture, with which the name of the species is
same species inhabiting a given environment. permanently associated. The type strain of a
Population dynamics: The study of factors species is marked by a superscript T at the end
affecting the variability of populations of of its identification number. The type strain is
microorganisms over time and space, includ- simply one of the first specimens of a
ing the interactions of these factors. described species. Unfortunately, many so-
Population genetics: The study of variation called type strains are in fact atypical species
in genes among a group of individual bacterial representatives.
strains, including the genetic evolution of Typing: Phenotypic and/or genetic analysis
populations. of bacterial isolates, below the species/subspe-
Selection: A natural process resulting in the cies level, performed in order to generate
evolution of an organism that is best adapted strain/clone-specific fingerprints or datasets
to a (selective) environment. that can be used, for example, to detect or rule
Species: The basic taxonomic category of out cross-infections, elucidate bacterial trans-
bacteria; a named group below the genus level mission patterns and find reservoirs or sources
whose members show a high degree of overall of infection in humans. ‘Subtyping’, a term
similarity as compared with other, more distantly commonly seen in American literature, is often
related, strains. There is currently no universally used as a synonym for typing.
accepted species definition in the context of Virulence: The property of an infectious
bacteriology, despite many attempts. agent that determines the extent to which an
Sporadic: Rare, occurring at unpatterned overt disease is produced in an infected pop-
irregular moments and localities, disconnected ulation.
in space and time; the opposite of epidemic
and endemic.
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–467
relevant to taxonomy, ecology and the study of
WHAT IS TYPING AND WHAT ARE
pathogenesis.
TYPING METHODS?
To put it simply, typing applies distinct labels to
Pathogenic bacteria replicate and persevere in bacterial isolates. These labels facilitate identifica-
ecological niches called reservoirs. Reservoirs tion of transmission routes and sources. However,
may be humans, including (fellow) patients and they can also contribute to in-depth investigations
healthcare personnel, animals, plants, water, of infectious disease pathogenesis, bacterial pop-
food and various niches in the environment. ulation structures and baterial genetics.
Transmission of bacteria from any of these Typing can be considered as either comparative
sources may generate clusters of colonisation or or definitive (library) typing. In comparative
infection among humans. Such clusters are typing, outbreak-related and unrelated isolates
recognised mostly as outbreaks of infectious are compared, since comparison of outbreak-
diseases. When these outbreaks are not con- related isolates with isolates from the past or the
trolled, major epidemics (due to unrestricted future is not relevant. This is sometimes consid-
further transmission) may arise. Bacterial epide- ered sufficient for outbreak investigation [20].
miological typing generates isolate-specific geno- However, in many outbreak settings, be they
typic or phenotypic characters that can be used nosocomial or community-based, it is often useful
to elucidate the sources and routes of spread of to compare strains from a current outbreak with
bacteria [46,47]. The scope of typing studies may previous strains, in which case a definitive
vary from purely ‘clinical’ (dissemination of (library) typing method should be used. There-
infections from patients, animals or other sources fore, it is important to set up and maintain
to non-colonised and uninfected individuals) to collections of alert organisms in any typing
‘environmental’ (the presence or spread of laboratory. Library systems are those that can be
organisms in inanimate surroundings) or even used in different laboratories, by different inves-
‘industrial’ (identification of organisms that are tigators at various time intervals, with the aim of
either valuable or a menace to bio-industry). generating high-quality data to be aggregated in a
Typing may also be used to identify emerging single database for comparative assessment, in
pathogenic strains or clones within a species, great detail at any time [51]. It is thus important
including potential agents of bioterrorism, in that the typing methods are robust and suffi-
forensic biology and as evidence in medico-legal ciently standardised to monitor the organisms of
cases. A variety of methods have been developed interest. While various multicentre studies aimed
to generate isolate-specific fingerprints, for epi- at standardising potential library typing methods
demiological typing. These methods should facil- have been undertaken with varying success, there
itate the determination of the relatedness among already exist a number of international networks
isolates derived from outbreak situations or incorporating databases compiled on the basis of
obvious and recent chains of transmission, in molecular typing data.
order to support or reject the hypothesis that the Typing can be undertaken at different levels,
isolates come from a single source. depending on the situation: locally, at a hospital
Typing data should always be considered or other primary laboratory, for small investiga-
within the time-frame and current epidemiolog- tions; regionally or nationally, in a reference
ical context that are being evaluated and from laboratory, to bear upon wider issues of public
which bacterial isolates have been obtained. For health and surveillance; or internationally
example, more variability can be expected through collaborative networks, to define or
between related isolates when longer time peri- survey the worldwide dissemination of major
ods are studied. The main focus of data inter- bacterial clones. At each of these levels, different
pretation in the clinical setting would be to methods may be applied.
identify sources, as opposed to reservoirs of
infection or colonisation [48–50]. Thus, typing
SETTING UP STRAIN COLLECTIONS
data can distinguish between cases linked to an
FOR TYPING LABORATORIES
outbreak of infections and those unrelated cases
due to more complex scenarios. In addition, The initiation and maintenance of strain collections
markers of biological diversity can also be are prerequisites for an epidemiological typing
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–468
study. The collection should comprise strains of the
Surveillance of infectious diseases
species of interest: epidemiologically unrelated
strains, sets of strains from outbreaks, and pro- Typing methods contribute useful information to
spective clinical isolates with well-defined inclu- epidemiological surveillance of infectious dis-
sion criteria. The number of strains and the eases, defined as a systematic, ongoing process
complexity of the collection are dependent upon of data collection, analysis, interpretation, dis-
the objective(s) of the research. The organisms semination of results, and action taken, aimed at
should be stored preferably in glycerol broth at recording disease trends and designing ways in
)80C or freeze-dried according to accepted guide- which to curb them [48,57–59]. Detection of
lines for strain preservation. Such collections are of clusters of defined pathogens (alert organisms)
much less value in the absence of a(n) (electronic) with a similar type may constitute an ‘early
database of relevant clinical, epidemiological and warning’ of a potential outbreak. Library typing,
demographical data concerning the strains at- such as serotyping, phage typing, PFGE or mul-
tached. Combining typing data with clinical and tilocus sequence typing (MLST), is mandatory for
demographical data is deemed to be extremely adequate surveillance of infectious diseases (for
important in deriving useful conclusions from examples, see Pitt [20]).
infectious diseases surveillance data. The com-
bined data should comprise: strain designation, Outbreak investigation
eventual other designations, species name, the
original specimen and its origin, date of isolation, An outbreak can be defined as a temporal
hospital, department, patient code, city, country, increase in the incidence of infection (or coloni-
and—for external strains—identity of provider. sation) by a certain bacterial species, caused by
Other relevant (optional) data are: antibiogram, enhanced transmission of a specific strain. It has
species identification method, and possible associ- to be noted that outbreaks can also be caused by
ation with an outbreak or otherwise. For strategic multiple strains. The increased occurrence of a
purposes, it is worthwhile to set up integrated single strain, therefore, needs to be distinguished
databases linking the hospital information system, from the fortuitous accumulation of sporadic
strain collection database and typing result data- cases. Nevertheless, while this holds true for
base, using appropriate software, either commer- healthcare-associated infections, it should be kept
cially acquired or developed in-house. in mind that in the case of foodborne infections,
for example, multi-strain outbreaks can also
occur. This situation is one of the many instances
REASONS FOR TYPING
when accurate epidemiological and clinical
Typing methods are used to study the spread descriptions are needed to prepare the design
and population dynamics of bacteria and other and corroborate the results of typing.
microorganisms in clinical and environmental In this context, typing methods are applied to
settings, at levels ranging from a single host to a generate and test hypotheses. Typing determines
global ecosystem. To date, these methods are the number of strains causing the increased
most easily and conveniently applied to haploid incidence and, ultimately, should help identify
organisms [40], but interest in the use of meth- the source(s) of contamination and the route of
ods for typing of diploid organisms, including transmission. Correct application of bacterial typ-
parasites, yeasts, fungi and plants, is growing ing will increase the efficacy of control measures
rapidly [52,53]. Finally, space (flight) microbiol- aimed at containing or interrupting the outbreak
ogy and the prevention of bioterrorism are new [60,61]. Unfortunately, the relevance of typing in
fields in which microbial typing is useful. In infection control strategies is still under-appreci-
forensic biology, nucleic acid technology is ated. Didactic instructions should, therefore, be
applied to human materials [54,55]. Interestingly, provided to those using typing in relation to
human forensics and microbial typing meet infection control [62–64]. This should lead to an
where bacteria can be used to collect criminal improved understanding of methodology and a
evidence or to scan crime scenes [56]. Finally, better overall appreciation of the added value of
genotypic methods can also be used in microbial epidemiological typing in the clinical setting. Cost
taxonomy. savings can be derived from curbing unnecessary
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–469
investigations or control measures when a sus- Technological aspects of the MLST method will be
pected outbreak is dismissed as an accumulation discussed in more detail in later sections of these
of sporadic cases derived from a single source. guidelines.
Study of pathogenesis and the course of CRITERIA FOR THE EVALUATION
infection AND VALIDATION OF
TYPING METHODS
We have already briefly mentioned the two major
uses of typing in studying infections affecting Before a typing method may be used in a given
more than one patient. However, typing can also situation, its appropriateness must have been
be used to elucidate the progress of infection in a clearly demonstrated. Every typing method there-
single patient, e.g., by differentiating between an fore needs to be evaluated and validated with
infection from endogenous microflora and that respect to a number of criteria [76–78]. These can
from an exogenous source [65]. When typing is be divided into performance and convenience
used to compare groups of strains that are either criteria. Because different investigations may
virulent or non-virulent, pathogenesis-related depend on different means and have different
markers can be identified. Such markers can requirements, there is no ideal, universally appli-
ultimately be translated into clinically relevant cable bacterial typing method [8]. Nevertheless,
diagnostic targets. the increasing need to communicate among labo-
ratories and to exchange outbreak investigation or
surveillance data requires some degree of agree-
Study of bacterial population genetics
ment on common methods. Such standardisation
Last but not least, some molecular typing systems is, of course, a lengthy and difficult process, but is
may be applied to large numbers of isolates from gradually being undertaken for the most popular
various origins in order to determine the intra- and dependable typing methods.
species population structure, and derive phylo-
genetic hypotheses from this structure [33–35,66].
Performance criteria
For example, PFGE analysis of the Pseudomonas
aeruginosa genome indicates that the average A good typing method should assess a marker
genomic pattern similarity of unrelated strains that remains stable during the study period, and
ranges between 20% and 60% with an average of does not vary to a degree that confuses the
35%, whereas clonally derived strains from a epidemiological picture. This marker should be
single host cluster at similarity levels above 80% testable in every isolate, i.e., it should provide
[66,67]. Similarly, high-resolution genomic finger- universal typeability of all isolates. It should also
printing of Acinetobacter has revealed that strains usefully discriminate among isolates, and this
of the same species cluster at 50% similarity or discrimination should be concordant with the
more, while the clone and strain delineation levels epidemiological picture. Finally, the results of a
are approximately 80% and 90%, respectively good typing method should be reproducible, inde-
[68–70]. pendently of the operator, place and time [79–81].
The current typing method of choice for A high degree of reproducibility will in turn make
performing bacterial population genetics studies, the results of the method amenable to inclusion in
and the one with the soundest biological basis, is databases and analysis by dedicated computer
MLST [71]. This sequence-based technique has software.
been applied to many important pathogens and
has provided valuable information concerning the Stability
evolution and diversification of these species. In This refers to the stability of the markers assessed
particular, these data have provided the means to by the typing method: a strain’s marker score
estimate how commonly bacterial genomes un- should not change rapidly and should correspond
dergo horizontal gene transfer and the impor- with the strain’s position in the epidemiological
tance that this process may have for the context. For example, the characteristics tested by a
emergence of clinically relevant strains with typing method should remain stable for each
heightened virulence or drug resistance [72–75]. isolate after its primary isolation and during
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–4610
laboratory storage and subculture. Preferably, the described, and nj is the number of strains
assessment of stability should also be performed in belonging to the jth type. The index should ide-
an in-vivo system. Although this may not always ally be 1.00 but, in practice, it should be at least in
be possible, successful examples have been the order of 0.95 for a typing system to be con-
reported in the literature [92]. Because mutations sidered more or less ‘ideal’. A 5% probability of
and recombination occur at frequencies dependent error is accepted by most professionals in the
upon species, strain and environmental condi- field. Calculations of the diversity index should
tions, the stability of the marker(s) tested by each be accompanied by critical assessment of the
method should be evaluated for each bacterial confidence interval, although this is very rarely
species studied [93,94]. Stability and reproducibil- done [87]. Typing methods exploring polymor-
ity (see below) are concepts that are sometimes phisms at multiple sites of the whole genome are
confused. To test stability, multiple subcultures of more likely to be more discriminatory than are
the same isolate, stored over different periods and methods exploring variation at a single locus. For
under different conditions, have to be processed in the purpose of calculation, non-typeable strains
the same run to minimise laboratory-introduced can be either excluded or grouped together,
variations [95]. A marker can also be considered to although the latter does not imply that they are of
be stable if multiple isolates of an epidemic strain the same type. In order to avoid overestimating
obtained from different patients at different the discriminatory power of a system, it is best
moments are indistinguishable by typing based that all untypeables be assembled into a single
on that particular marker. group.
Typeability Epidemiological concordance
This refers to a method’s ability to assign a type to The results of a typing method should reflect,
all isolates tested by it. It can be expressed as the agree with, and possibly further illuminate the
percentage of typeable isolates over the total available epidemiological information about the
number of typed (typeable and non-typeable) cases of colonisation or infection under study.
isolates [82–84]. Whereas most of the genotyping For example, epidemiologically related isolates
methods can characterise all of the isolates within derived from presumably single-strain or single-
a population (100% typeability), typeability can clone outbreaks should be assigned to identical or
be low with classic phenotypic methods such as related types [22,23]. When validating a method,
serotyping, due to the fact that the existing it is desirable that several sets, e.g., five or more,
serotyping schemes do not cover genetic variation of outbreak-related strains (n = five to ten isolates
in full. per set) are included in the test population (see
below). Phenotypic methods are usually less
Discriminatory power likely to be concordant with epidemiology when,
This refers to a method’s ability to assign a for example, distinct strains display similar phe-
different type to two unrelated strains sampled notypes (due to evolutionary convergence) [96].
randomly from the population of a given species.
It can be expressed as a probability using Simp- Reproducibility
son’s index of diversity [85,86]. Hunter and This refers to the ability of a typing method to
Gaston’s modification of Simpson’s index of assign the same type to an isolate tested on
diversity and fixed confidence intervals are independent occasions, separated in time and/or
important parameters used for making a decision place [88]. The reproducibility of a marker pattern
on strain identity or diversity [86]. The formula (or data generation in general) and that of type
used to define the diversity index or, better, assignment (data interpretation) may be different,
Simpson’s index of diversity D is: and both need to be evaluated. Reproducibility
may be influenced by many steps in a procedure,
1 XS
D¼1 nj ðnj 1Þ; as a result of either the protocol used or the
NðN 1Þ J¼1 stringency of its application. Factors to consider
include: the preparation of materials (e.g., varia-
where N is the total number of strains in the tion in growth conditions, and methods of DNA
sample population, S is the total number of types extraction), different batches or reagents, or
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–4611
reagent variation as a result of local preparation, published more than 10 years ago, it has not
different types of equipment, bias in observing been adopted very widely. Publications in which
and recording the results, and, finally, analysis appropriate test populations are analysed in
and interpretation of results. Reproducibility has detail are rare, and the mathematics required to
both intra-laboratory and inter-laboratory dimen- support the corresponding conclusions are
sions. Both require standardised protocols and hardly ever applied.
adequate personnel training to ensure a reliable
method that produces results that are ‘fit for
Convenience criteria
purpose’ for different organisms in different
settings [89–91]. Once the intrinsic value of a method, as well as its
appropriateness for the typing of a specific spe-
Test population cies, has been established on the basis of the
An appropriate and well-defined test population performance criteria discussed above, another set
is a prerequisite for evaluating the typeability, of criteria, those related to feasibility or conve-
discriminatory power and epidemiological con- nience, need to be considered. These are impor-
cordance of typing methods. Note that the tant for the selection of an appropriate typing
nature of such a population is, of course, defined method, depending on a number of factors, such
by the epidemiological context, the species of as the scale of the investigation, the timeliness
organism involved, whether the studies are local, required of the results, and the financial and
regional or global, and whether long-term sur- technical resources available. The following crite-
veillance is required. A large test population of ria of convenience, therefore, need to be consid-
isolates correctly identified to the species level ered: flexibility, rapidity, accessibility, ease of use,
(preferably n > 100) should be assembled to costs, and suitability for computerised analysis
reflect as much as possible the diversity expected and storage of results [97]. The portability of
in the species as a whole, or at least in the sub- results is being improved continuously, and this
population to which the typing method will be latter criterion is becoming increasingly impor-
applied [20–23]. It is recommended to cover as tant.
many ecological niches as may be included in
future investigations, such as particular patient Flexibility (or spectrum)
populations (including age category, immune This reflects the range of species that are typeable
status, type of hospital and ward, geographical with minimal modifications of the method [98].
origin) and relevant environmental reservoirs The broader the range of bacterial species that can
(e.g., for zoonoses or foodborne and waterborne be studied, the more central the position of the
infections). The test population should include method in the general typing laboratory will be.
strains that are presumably unrelated epidemio- Modern DNA sequence-based methods show
logically, on the basis of detailed clinical and optimal flexibility in the sense that the principle,
epidemiological data, as well as outbreak-related as well as the skills and equipment required, are
isolates. For these reasons, it is important that the same for different species. Nevertheless, these
hospital epidemiologists invest in prospective methods still need to be optimised and validated
collections of organisms that have given rise to for each species of interest; e.g., amplification
important healthcare-associated outbreaks. The primers developed for one species are usually not
test population is distinct from the panels of useful for another.
control isolates that should be used in many
studies. For example, in outbreak investigations, Rapidity
the appropriate level of discrimination of the This refers to the total time required to get from
typing method(s) should be confirmed by com- the bacterial isolates to the final typing results.
paring the outbreak-related strains to a set of The highest degree of typing rapidity can be
control strains (n = 10–30) from a similar time attained with methods that are applied directly to
period, locality and patient population, but clinical materials, the so-called culture-indepen-
which are, a priori, not epidemiologically related. dent procedures [99,100]. Ideally, typing should
We feel compelled to emphasise that, although be performed in ‘real time’; having results avail-
the earlier version of the current guidelines was able within a single working day would strongly
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–4612
enhance the clinical impact of epidemiological tutes to monitor the spread of such strains or
typing in general medicine. clones beyond the hospital level. Of course,
computerised analysis is optimal in combination
Accessibility with library methods of typing, with MLST as
This depends upon the availability of reagents the current key example.
and equipment, as well as the skills required for a
given method in a given laboratory.
VALIDATION OF NEW METHOD–
MICROBE COMBINATIONS
Ease of use
This encompasses technical simplicity, workload, Application of any typing method requires care-
suitability for processing large numbers of iso- ful assessment of its suitability for a species not
lates, and ease of scoring and interpreting the yet analysed by it. New methods or variants of
results. existing ones are published on a regular basis
[102], but they vary widely in terms of how well
Cost validated they are. It cannot be emphasised
This depends on numerous factors. For example, enough that testing limited numbers of bacterial
there is the amount of the initial capital outlay for isolates without adequate follow-up, using non-
the equipment, its depreciation, which will validated technology in merely local applications,
depend on whether it is out-of-date compared should be discouraged. In the current era, when
with newer versions or totally new platforms, the complete genome sequences are available for
frequency and care with which it is used, and multiple strains of most, if not all, clinically
finally, the costs of any modifications to the room. relevant microorganisms, such sequence deposi-
The latter could include the additional options of tories can generate important clues for the selec-
extra air-conditioning and floor reinforcement. tion of appropriate molecular typing targets.
The costs of servicing, the price, need for and Protocols for frequently used typing methods
ready availability of replacement parts, and the should be validated according to the recommen-
cost of consumable reagents should also be dations given in this article by networks of expert
considered. Then there are staffing costs, which laboratories. Subsequently, certified ‘end-user’
will depend on the time required to perform laboratories should attentively adhere to these
procedures, the number and grade of personnel protocols. Admittedly, the latter simple statement
required, their training and requirements for is often difficult to translate into practice; the
demonstration of competencies for accreditation personal preferences of many scientists can
or other purposes. These costs can be offset, for severely compromise the objective of working
example, by income generation, which will according to a standardised protocol. In conclu-
depend on the ability to provide typing services sion, inter-method validation is important and
for others or income-generating training courses necessary, both from a theoretical point of view
for others to learn the typing method. and from a practical perspective [103].
Amenability to computerised analysis and
PRINCIPLES AND OVERVIEW OF
incorporation of typing results in electronic databases
CURRENT TYPING METHODS
These two factors are most important for longi-
tudinal comparison of large numbers of isolates. Over the past two decades, a plethora of novel
At the local (hospital) level, data obtained by and often innovative typing methods has been
robust typing methods can be analysed elec- developed. These range from methods that assess
tronically or assessed visually. Visual interpre- simple phenotypic traits to DNA sequencing.
tation, even when only small numbers of Previously, the comparison of phenotypic char-
isolates are studied, requires normalisation of acters, which involves the comparison of appar-
the data prior to inspection [101]. Nevertheless, ent biological features of isolates, was often
since clones are spreading among hospitals or in abandoned because of the problems with perfor-
the community, both regionally and globally, it mance criteria already mentioned. Instead, meth-
is important that electronic databases be created, ods involving the comparison of genomic DNA
enabling microbiologists and public health insti- fragments were adopted. DNA molecules (or
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–4613
restriction fragments or amplified sections there- substances (be they metabolites, drugs, bacterial
of) can be separated on the basis of their molec- toxins or bacteriophages) and their expression of
ular size by gel electrophoresis. Such size specific molecules (be they surface antigens or
comparisons assess differences in the length of allelic variants of housekeeping enzymes). All
DNA fragments obtained from DNA from differ- methods require strict standardisation of experi-
ent bacterial strains. Whether the fragments of mental conditions, since phenotypes are generally
DNA are natural (e.g., plasmids) or generated at quite susceptible to changes in environmental
random, by restriction enzymes or after amplifi- conditions. In a simple statement: phenotyping
cation of the DNA using enzymatic DNA repli- results in the grouping of organisms according to
cation (PCR), does not matter; size differences, their similarity in characters resulting from the
provided that they are accurately determined, can expression of their genotypes.
be excellent markers of strain differences. Biotyping assesses biochemical characteristics
By definition, the genome of every bacterial that are known to vary within a given species.
isolate is unique. The mere fact that DNA Typeability is usually excellent. Discriminatory
polymerases make copying mistakes during rep- power is variable and, to optimise it, a large
lication suggests that no genome has a 100% number of well-selected characteristics, e.g., meta-
identical counterpart [104]. However, such muta- bolic reactions, needs to be included in the test
tions must be compatible with nature; they must scheme. Stability is dependent on the species and
be neutral or at least in line with existing characteristic under consideration. The methods
structure–function relationships among the corre- are usually technically easy and inexpensive, the
sponding gene products. Hence, bacterial strains data generated are simple to score and interpret,
differ with respect to their complete genome and all tests can be performed, even in the
sequence, and DNA sequencing methodologies smallest of laboratories, on large numbers of
can therefore be used to assess similarity of isolates. If reproducibility is demonstrated, it can
strains. A challenge for the near future is to be used as a library typing method [109,110]. For
assess which DNA sequences are useful epidemi- instance, commercial systems facilitating the mea-
ological markers, a task that is greatly assisted by surement of large panels of ‘biotype characteris-
whole genome sequencing [105–107]. tics’ have been developed. These systems use
Since far more detailed reviews exist concern- versatile redox technologies, enabling the quanti-
ing the technical aspects of typing methods fication of various biochemical reactions by colour
[50,108], we will restrict ourselves to defining readings [111–114]. The main power of the system
briefly the common aspects and quality charac- lies in its ability to distinguish among strains
teristics of the methods, without any claim to within a species [115,116]. Phenotype reaction
completeness. The diversity and plethora of arrays are available and are useful tools in
methods available to the scientific community addition to DNA and proteomic technologies.
are such that it is impossible to be comprehensive The reproducibility of biotyping is organism- and
in the subsequent sections. Strategic literature character-dependent. It is rarely 100%.
references will be included to facilitate and Antimicrobial susceptibility testing (antibio-
stimulate further reading. Important overviews gram-based typing) can be performed either by
of typing methods can also be found in several drug diffusion in solid growth media or drug
general textbooks on the practical and theoretical dilution in liquid media using a variety of
aspects of bacterial typing. measurement systems. Most clinical microbiology
laboratories perform some sort of antibiogram
typing, since its results are commonly used to
Phenotypic typing methods
guide chemotherapy. Therefore, this method has
Phenotyping may involve colony morphology, immediate clinical consequences also. Antibio-
colour, odour and other macroscopic features, gram-based typing can, with appropriate selec-
but most typing methods rely on traits that tion of drugs, be applied to most species.
require specialised technology in order to be Discrimination is dependent on the diversity,
documented. For example, they may assess, stability and relative prevalence of the detect-
qualitatively and quantitatively, the ability of able acquired resistance mechanisms in study
isolates to grow in the presence of specific isolates. It is also dependent on the number of
2007 The Authors
Journal compilation 2007 Clinical Microbiology and Infectious Diseases, CMI, 13 (Suppl. 3), 1–46You can also read
