Establishing a many-cytokine signature via multivariate anomaly detection.

Establishing a cytokine signature associated to some medical condition is an important task in immunology. Increasingly, large numbers of cytokines are used for signatures, via lists of reference ranges for each individual cytokine or ratios of cytokines. Here we argue that this common approach has weaknesses, especially when many different cytokines are analysed. Instead, we propose that establishing signatures can be framed as a multivariate anomaly detection problem, and hence exploit the many statistical methods available for this. In this framework, whether or not a given subject's profile matches the cytokine signature of some condition is determined by whether or not the profile is typical of reference samples of that condition, as judged by an anomaly detection algorithm. We examine previously published cytokine data sets associated to pregnancy complications, brain tumours, and rheumatoid arthritis, as well as normal healthy control samples, and test the performance of a range of anomaly detection algorithms on these data, identifying the best performing methods. Finally, we suggest that this anomaly detection approach could be adopted more widely for general multi-biomarker signatures.

Longitudinal impact of the Women's Health Initiative study on hormone therapy use in Australia.

Purpose: This study aimed to examine the longitudinal impact of evidence changes on menopausal hormone therapy (MHT) use in Australia. Methods: We analyzed two datasets of subsidized and total MHT use (2000-2016) using segmented regression analysis to explore the impact of the Women's Health Initiative (WHI) 2002 and 2007 studies. Analyses were stratified by class, route, and strength. Use was measured in defined daily dose/1000 women/day (DDD/1000/day) or packs/1000 women/month (packs/1000/month). Results: The drop in total MHT use after the WHI 2002 was substantial. The biggest decreases in class, route, and strength were estrogens (28.99 DDD/1000/day, 95% confidence interval [CI] 23.97, 34.01), oral (46.07 DDD/1000/day, 95% CI 41.13, 51.01), and medium strength (34.95 packs/1000/month, 95% CI 30.17, 39.73), respectively. However, vaginal use remained stable (-1.83 DDD/1000/day, 95% CI -3.83, 0.17). Profiles of total and subsidized use were similar over time. Utilization levels were relatively unchanged after 2007. Decreased utilization contributed to product discontinuation, with a lag of up to 4 years. Product discontinuation in 2009 further decreased utilization. Discussion and conclusions: MHT use remained low after 2002 despite evidence favoring its use in women younger than 60 years or within 10 years postmenopause. Continued low use could relate to the WHI 2002 media coverage, therapy objectives, key stakeholder uncertainty, health policies, and medicine availability.

Emergence and spread of predominantly community-onset Clostridium difficile PCR ribotype 244 infection in Australia, 2010 to 2012.

We describe an Australia-wide Clostridium difficile outbreak in 2011 and 2012 involving the previously uncommon ribotype 244. In Western Australia, 14 of 25 cases were community-associated, 11 were detected in patients younger than 65 years, 14 presented to emergency/outpatient departments, and 14 to non-tertiary/community hospitals. Using whole genome sequencing, we confirm ribotype 244 is from the same C. difficile clade as the epidemic ribotype 027. Like ribotype 027, it produces toxins A, B, and binary toxin, however it is fluoroquinolone-susceptible and thousands of single nucleotide variants distinct from ribotype 027. Fifteen outbreak isolates from across Australia were sequenced. Despite their geographic separation, all were genetically highly related without evidence of geographic clustering, consistent with a point source, for example affecting the national food chain. Comparison with reference laboratory strains revealed the outbreak clone shared a common ancestor with isolates from the United States and United Kingdom (UK). A strain obtained in the UK was phylogenetically related to our outbreak. Follow-up of that case revealed the patient had recently returned from Australia. Our data demonstrate new C. difficile strains are an on-going threat, with potential for rapid spread. Active surveillance is needed to identify and control emerging lineages.

Typing of Campylobacter jejuni isolates from dogs by use of multilocus sequence typing and pulsed-field gel electrophoresis.

Campylobacter is a major cause of human gastroenteritis worldwide. Risk of Campylobacter infection in humans has been associated with many sources, including dogs. This study aimed to investigate whether C. jejuni carried by dogs could potentially be a zoonotic risk for humans and if there were common sources of C. jejuni infection for both humans and dogs. Multilocus sequence typing (MLST) together with macrorestriction analysis of genomic DNA using SmaI and pulsed-field gel electrophoresis (PFGE) were both used to analyze 33 C. jejuni isolates obtained from various dog populations, including those visiting veterinary practices and from different types of kennels. MLST data suggested that there was a large amount of genetic diversity between dog isolates and that the majority of sequence types found in isolates from these dogs were the same as those found in isolates from humans. The main exception was ST-2772, which was isolated from four samples and could not be assigned to a clonal complex. The most commonly identified clonal complex was ST-45 (11 isolates), followed by ST-21 (4 isolates), ST-508 (4 isolates), and ST-403 (3 isolates). The profiles obtained by macrorestriction PFGE were largely in concordance with the MLST results, with a similar amount of genetic diversity found. The diversity of sequence types found within dogs suggests they are exposed to various sources of C. jejuni infection. The similarity of these sequence types to C. jejuni isolates from humans suggests there may be common sources of infection for both dogs and humans. Although only a small number of household dogs may carry C. jejuni, infected dogs should still be considered a potential zoonotic risk to humans, particularly if the dogs originate from kennelled or hunt kennel dog populations, where the prevalence may be higher.

Genetic diversity and stability of the porA allele as a genetic marker in human Campylobacter infection.

The major outer-membrane protein (MOMP) of Campylobacter jejuni and Campylobacter coli, encoded by the porA gene, is extremely genetically diverse. Conformational MOMP epitopes are important in host immunity, and variation in surface-exposed regions probably occurs as a result of positive immune selection during infection. porA diversity has been exploited in genotyping studies using highly discriminatory nucleotide sequences to identify potentially epidemiologically linked cases of human campylobacteriosis. To understand the overall nature and extent of porA diversity and stability in C. jejuni and C. coli we investigated sequences in isolates (n=584) obtained from a defined human population (approx. 600,000) over a defined time period (1 year). A total of 196 distinct porA variants were identified. Regions encoding putative extracellular loops were the most variable in both nucleotide sequence and length. Phylogenetic analysis identified three porA allele clusters that originated in (i) predominantly C. jejuni and a few C. coli, (ii) solely C. jejuni or (iii) predominantly C. coli and a few C. jejuni. The stability of porA within an individual human host was investigated using isolates cultured longitudinally from 64 sporadic cases, 27 of which had prolonged infection lasting between 5 and 98 days (the remainder having illness of normal duration, 0-4 days), and 20 cases from family outbreaks. Evidence of mutation was detected in two patients with prolonged illness. Despite demonstrable positive immune selection in these two unusual cases, the persistence of numerous variants within the population indicated that the porA allele is a valuable tool for use in extended typing schemes.

Comparison of Campylobacter populations in wild geese with those in starlings and free-range poultry on the same farm.

Wild geese are a potential source of Campylobacter infection for humans and farm animals and have been implicated in at least two large waterborne disease outbreaks. There have been few investigations into the population biology of Campylobacter in geese, carriage rates are reported to vary (0 to 100%), and no genetic characterization of isolates has been performed. Fecal samples collected from wild geese in Oxfordshire, United Kingdom, were culture positive for C. jejuni (50.2%) and C. coli (0.3%). The C. jejuni (n = 166) isolates were characterized by using multilocus sequence typing and were compared with isolates collected from free-range broiler chickens and wild starlings sampled at the same location. A total of 38 STs, six clonal complexes, and 23 flaA SVR nucleotide STs were identified. The ST-21 and ST-45 complexes (5.4% of isolates) were the only complexes to be identified among isolates from the geese and the other bird species sampled in the same location. These clonal complexes were also identified among human disease isolates collected in the same health care region. The results indicate that large numbers of wild geese carry Campylobacter; however, there was limited mixing of Campylobacter populations among the different sources examined, and the host source could be predicted with high probability from the allelic profile of a C. jejuni isolate. In conclusion, genotypes of C. jejuni isolated from geese are highly host specific, and a comparison with isolates from Oxfordshire cases of human disease revealed that while geese cannot be excluded as a source of infection for humans and farm animals, their contribution is likely to be minor.

Use of multilocus sequence typing to investigate the association between the presence of Campylobacter spp. in broiler drinking water and Campylobacter colonization in broilers.

The presence of campylobacters in broiler chickens and throughout the broiler water delivery systems of 12 farms in northeastern Scotland was investigated by sensitive enrichment methods and large-volume filtration. Campylobacter presence was independent of the water source and whether the water was treated. The genotypes of Campylobacter jejuni isolates recovered from chickens and various locations within the water delivery systems were compared by multilocus sequence typing. Matching strains in shed header tanks and birds were found at 1 of the 12 farms investigated. However, the sequence of contamination or whether the source was within or outside the shed was not determined. Nevertheless, these data provide evidence that drinking water could be associated with broiler infection by campylobacters.

Multilocus sequence typing of Campylobacter jejuni isolates from New South Wales, Australia.

AIMS: Multilocus sequence typing (MLST) was used to examine the diversity and population structure of Campylobacter jejuni isolates associated with sporadic cases of gastroenteritis in Australia, and to compare these isolates with those from elsewhere. METHODS AND RESULTS: A total of 153 Camp. jejuni isolates were genotyped. Forty sequence types (STs) were found, 19 of which were previously undescribed and 21 identified in other countries. The 19 newly described STs accounted for 43% of isolates, 16 of which were assigned to known clonal complexes. Eighty-eight percent of isolates were assigned to a total of 15 clonal complexes. Of these, four clonal complexes accounted for 60% of isolates. Three STs accounted for nearly 40% of all isolates and appeared to be endemic, while 21 STs were represented by more than one isolate. Seven infections were acquired during international travel, and the associated isolates all had different STs, three of which were exclusive to the travel-acquired cases. Comparison of serotypes among isolates from clonal complexes revealed further diversity. Eight serotypes were identified among isolates from more than one clonal complex, while isolates from six clonal complexes displayed serotypes not previously associated with those clonal complexes. CONCLUSIONS: Multilocus sequence typing is a useful tool for the discrimination of subtypes and examination of the population structure of Camp. jejuni associated with sporadic infections. SIGNIFICANCE AND IMPACT OF THE STUDY: This study highlights the genotypic diversity of Camp. jejuni in Australia, demonstrating that STs causing disease have both a global and a local distribution evident from the typing of domestically and internationally acquired Camp. jejuni isolates.

Reference isolates for the clonal complexes of Campylobacter jejuni.

AIMS: To identify and make available through the National Collection of Type Cultures (NCTC) a set of reference isolates for the clonal complexes of Campylobacter jejuni. METHODS AND RESULTS: The development of a multilocus sequence typing scheme for C. jejuni enabled the genetic characterization of a large number of isolates (n = 814) from cases of human disease, animals, birds and their food products. The nucleotide sequence data were used to assign each isolate an allelic profile or sequence type (ST) and examine the C. jejuni population structure in terms of clonal complexes. The clonal complexes consisted of an abundant central or founder genotype (ST), after which the complex was named, together with very closely related, generally less abundant genotypes differing from the founder at one, two or three loci. The clonal complex is an informative unit for the study C. jejuni epidemiology. It provides data which enabled the choice of 13 C. jejuni founder isolates for submission to the NCTC as a representative cross-section of the C. jejuni population. CONCLUSIONS: These 13 isolates provide a defined resource for further research into aspects of C. jejuni biology such as genomic diversity, virulence and adaptation to particular hosts or environmental survival. SIGNIFICANCE AND IMPACT OF STUDY: This isolate collection is available through the NCTC and provides a resource for further research.

Sequence typing confirms that Campylobacter jejuni strains associated with Guillain-Barré and Miller-Fisher syndromes are of diverse genetic lineage, serotype, and flagella type.

Guillain-Barré syndrome (GBS) and Miller-Fisher syndrome (MFS) are correlated with prior infection by Campylobacter jejuni in up to 40% of cases. Nucleotide sequence-based typing of 25 C. jejuni isolates associated with neuropathy permitted robust comparisons with equivalent data from approximately 800 C. jejuni isolates not associated with neuropathy. A total of 13 genetic lineages and 20 flaA short variable region nucleotide sequences were present among the 25 isolates. A minority of isolates (4 of 25) had the flaA short variable region nucleotide sequences that were previously proposed as a marker for GBS-associated isolates. These 4 isolates probably represented the Penner serotype 19 lineage, which has been proposed to have an association with GBS.

Multilocus sequence typing system for Campylobacter jejuni.

The gram-negative bacterium Campylobacter jejuni has extensive reservoirs in livestock and the environment and is a frequent cause of gastroenteritis in humans. To date, the lack of (i) methods suitable for population genetic analysis and (ii) a universally accepted nomenclature has hindered studies of the epidemiology and population biology of this organism. Here, a multilocus sequence typing (MLST) system for this organism is described, which exploits the genetic variation present in seven housekeeping loci to determine the genetic relationships among isolates. The MLST system was established using 194 C. jejuni isolates of diverse origins, from humans, animals, and the environment. The allelic profiles, or sequence types (STs), of these isolates were deposited on the Internet (http://mlst.zoo.ox.ac.uk), forming a virtual isolate collection which could be continually expanded. These data indicated that C. jejuni is genetically diverse, with a weakly clonal population structure, and that intra- and interspecies horizontal genetic exchange was common. Of the 155 STs observed, 51 (26% of the isolate collection) were unique, with the remainder of the collection being categorized into 11 lineages or clonal complexes of related STs with between 2 and 56 members. In some cases membership in a given lineage or ST correlated with the possession of a particular Penner HS serotype. Application of this approach to further isolate collections will enable an integrated global picture of C. jejuni epidemiology to be established and will permit more detailed studies of the population genetics of this organism.

Interactions between hepatitis delta virus proteins.

The 195- and 214-amino-acid (aa) forms of the delta protein (deltaAg-S and deltaAg-L, respectively) of hepatitis delta virus (HDV) differ only in the 19-aa C-terminal extension unique to deltaAg-L. deltaAg-S is needed for genome replication, while deltaAg-L is needed for particle assembly. These proteins share a region at aa 12 to 60, which mediates protein-protein interactions essential for HDV replication. H. Zuccola et al. (Structure 6:821-830, 1998) reported a crystal structure for a peptide spanning this region which demonstrates an antiparallel coiled-coil dimer interaction with the potential to form tetramers of dimers. Our studies tested whether predictions based on this structure could be extrapolated to conditions where the peptide was replaced by full-length deltaAg-S or deltaAg-L, and when the assays were not in vitro but in vivo. Nine amino acids that are conserved between several isolates of HDV and predicted to be important in multimerization were mutated to alanine on both deltaAg-S and deltaAg-L. We found that the predicted hierarchy of importance of these nine mutations correlated to a significant extent with the observed in vivo effects on the ability of these proteins to (i) support in trans the replication of the HDV genome when expressed on deltaAg-S and (ii) act as dominant-negative inhibitors of replication when expressed on deltaAg-L. We thus infer that these biological activities of deltaAg depend on ordered protein-protein interactions.

Characterization of the 5' ends for polyadenylated RNAs synthesized during the replication of hepatitis delta virus.

The genome of hepatitis delta virus (HDV) is a 1,679-nucleotide (nt) single-stranded circular RNA that is predicted to fold into an unbranched rodlike structure. During replication, two complementary RNAs are also detected: an exact complement, referred to as the antigenome, and an 800-nt polyadenylated RNA that could act as the mRNA for the delta antigen. We used a 5' rapid amplification of cDNA ends procedure, followed by cloning and sequencing, to determine the 5' ends of the polyadenylated RNAs produced during HDV genome replication following initiation under different experimental conditions. The analyzed RNAs were from the liver of an infected woodchuck and from a liver cell line at 6 days after transfection with either an HDV cDNA or ribonucleoprotein (RNP) complexes assembled in vitro with HDV genomic RNA and purified recombinant small delta protein. In all three situations the 5' ends mapped specifically to nt 1630. In relationship to what is called the top end of the unbranched rodlike structure predicted for the genomic RNA template, this site is located 10 nt from the top, and in the middle of a 3-nt external bulge. Following transfection with RNP, such specific 5' ends could be detected as early as 24 h. We next constructed a series of mutants of this predicted bulge region and of an adjacent 6-bp stem and the top 5-nt loop. Some of these mutations decreased the ability of the genome to undergo antigenomic RNA synthesis and accumulation and/or altered the location of the detected 5' ends. The observed end located at nt 1630, and most of the novel 5' ends, were consistent with transcription initiation events that preferentially used a purine. The present studies do not prove that the detected 5' ends correspond to initiation sites and do not establish the hypothesis that there is a promoter element in the vicinity, but they do show that the location of the observed 5' ends could be controlled by nucleotide sequences at and around nt 1630.

The seroepidemiology of genogroup 1 and genogroup 2 Norwalk-like viruses in Italy.

Southampton virus (SV) and Lordsdale viruses (LV) are small round structured viruses characterised recently and belong to two separate genogroups. The capsid genes of these viruses were expressed in insect cells using recombinant baculoviruses. Both SV (genogroup 1) and LV (genogroup 2) capsid proteins self-assembled to form virus-like particles (VLPs). The VLPs were used in a standard enzyme-linked immunosorbent assay (ELISA) to screen for antibodies to SV and LV in 1,729 age-stratified human sera collected in Verona, Italy between January and November 1996. SV VLPs were labile compared with LV VLPs. There was a large difference in the prevalence of SV (28.7%) compared with LV (91.2%). However, presentation of SV VLPs using chicken egg yolk antibody-coated wells (IgY capture ELISA) with a subset of serum samples from patients (0-19 years) increased the number of positive sera significantly (50.5%), indicating that SV antigen integrity is an important factor in the assay. Recent reverse transcription-polymerase chain reaction (RT-PCR) studies have shown that LV is circulating currently and analysis of IgY capture ELISA data showed greater reactivity for LV than SV, reflecting a genuinely lower rate of recent infection by this genogroup 1 virus.

Electrophoretic analysis of the ribonucleoproteins of hepatitis delta virus.

Replication of hepatitis delta virus (HDV) is dependent on delta antigen (deltaAg), an HDV-encoded protein, which binds to HDV RNA and is capable of multimerization. To characterize HDV-specific ribonucleoprotein complexes (RNP) we used electrophoresis into non-denaturing agarose gels followed by northern analysis, to detect HDV RNA, and immunoblot, to detect deltaAg. We studied RNP from three sources: (i) vRNP, disrupted virions obtained from infected woodchuck serum; (ii) sRNP, disrupted particles secreted from transfected cultured cells; and (iii) cRNP, isolated from cells in which HDV genome replication was occurring. sRNP were approximately 28% smaller than vRNP. Treatment of vRNP with aurin tricarboxylic acid disrupted both deltaAg-deltaAg and deltaAg-RNA interactions while vanadyl ribonucleosides released the RNA without causing detectable disruption of the multimeric deltaAg complex. cRNP were smaller and more heterogeneous than vRNP and sRNP, and probably contained host components. The application of these electrophoretic procedures, and especially the use of prior treatments with vanadyl ribonucleoside complexes have provided valuable information on the RNP of HDV, and we expect they should find applicability in RNP studies of other RNA viruses.

Initiation of hepatitis delta virus genome replication.

The small, 195-amino-acid form of the hepatitis delta virus (HDV) antigen (deltaAg-S) is essential for genome replication, i.e., for the transcription, processing, and accumulation of HDV RNAs. To better understand this requirement, we used purified recombinant deltaAg-S and HDV RNA synthesized in vitro to assemble high-molecular-weight ribonucleoprotein (RNP) structures. After transfection of these RNPs into human cells, we detected HDV genome replication, as assayed by Northern analysis or immunofluorescence microscopy. Our interpretation is that the input deltaAg-S is necessary for the RNA to undergo limited amounts of RNA-directed RNA synthesis, RNA processing, and mRNA formation, leading to de novo translation of deltaAg-S. It is this second source of deltaAg-S which then goes on to support genome replication. This assay made it possible to manipulate in vitro the composition of the RNP and then test in vivo the ability of the complex to initiate RNA-directed RNA synthesis and go on to achieve genome replication. For example, both genomic and antigenomic linear RNAs were acceptable. Substitution for deltaAg-S with truncated or modified forms of the deltaAg, and even with HIV nucleocapsid protein and polylysine, was unacceptable; the exception was a form of deltaAg-S with six histidines added at the C terminus. We expect that further in vitro modifications of these RNP complexes should help define the in vivo requirements for what we define as the initiation of HDV genome replication.

Correlation of patient immune responses with genetically characterized small round-structured viruses involved in outbreaks of nonbacterial acute gastroenteritis in the United States, 1990 to 1995.

Small round-structured viruses (SRSVs) are a genetically and antigenically diverse group of caliciviruses that are the most common cause of outbreaks of acute nonbacterial gastroenteritis. We have applied both molecular techniques to characterize SRSVs in fecal specimens and serologic assays using four different expressed SRSV antigens to examine the distribution of outbreak strains in the United States and determine if the immune responses of patients were strain specific. Strains from 23 outbreaks of SRSV gastroenteritis were characterized by reverse transcription-PCR and nucleotide sequencing of a 277-base region of the capsid gene. These strains segregated into two distinct genogroups, I and II, comprising four and six clusters of strains respectively, each representing a distinct phylogenetic lineage. Serum IgG responses in patients were measured by enzyme immunoassay using expressed capsid antigens of Norwalk virus (NV), Toronto virus (TV), Hawaii virus (HV), and Lordsdale virus (LV), representing four of the 10 clusters. While strains in genogroups I and II were antigenically distinct, within genogroups, the specificity of the immune response varied greatly. Patients infected with genogroup I strains which had as much as 38.5% aa divergence from NV demonstrated relatively homologous seroresponses to the single NV antigen. In contrast, in genogroup II, homologous seroresponses to TV and HV were only present when the infecting strains showed less than 6.5% aa divergence from these antigens. These results suggest that TV and HV represent not only separate genetic clusters in genogroup II but also separate antigenic groups, each of which is related but distinguishable. In addition, two genetically distinct SRSV strains were identified for which we have no homologous antigen. This study suggests that while current molecular diagnostics are capable of detecting the full range of SRSVs, additional expressed antigens will be required to detect an immune response to SRSV infection caused by all the antigenically diverse strains.

Human enteric Caliciviridae: the complete genome sequence and expression of virus-like particles from a genetic group II small round structured virus.

Comparisons of the RNA polymerase and capsid sequences of small round structured viruses (SRSVs) have recently shown these are genetically diverse viruses which fall into two distinct groups. The genomes of two group I viruses, Southampton and Norwalk viruses have been characterized; however, similar data for the genetic group II SRSVs have not been available until now. We report here the complete genome sequence of a recent group II SRSV, Lordsdale virus. The Lordsdale virus genome is 7555 nt in length and has a similar organization to the group I SRSVs. The large ORF in the 5' half of the genome (5100 nt) is shorter than the group I SRSV ORF1 (5367 nt), but has the characteristic 2C helicase, 3C protease and 3D RNA polymerase enzyme motifs. ORF2, encoding the structural protein is of a similar size to the group I viruses but the small 3'-terminal ORF is significantly larger in group II. A highly conserved sequence of 28 nt was identified at the start of Lordsdale virus ORF1 and repeated at the start of ORF2. These conserved motifs are typical of the animal caliciviruses. Comparison of the 150 N-terminal amino acids in the ORF1 protein revealed little identity between the two SRSV genetic groups, reflecting the shorter ORF1 in the group II virus. Recombinant baculoviruses containing ORF2 and ORF3 sequences were constructed and used to express large quantities of the group II Lordsdale virus structural protein. The capsid protein formed virus-like particles by self assembly which resembled 'empty' SRSVs.