Article Text

Original article
Use of Chlamydia trachomatis high-resolution typing: an extended case study to distinguish recurrent or persistent infection from new infection
  1. Hannelore M Götz1,2,
  2. Reinier J M Bom3,
  3. Mireille E G Wolfers1,
  4. Johan Fennema4,
  5. Ingrid V F van den Broek5,
  6. Arjen G C L Speksnijder3,
  7. Sylvia M Bruisten3
  1. 1Division of Infectious Disease Control, Rotterdam-Rijnmond Public Health Service, Rotterdam, The Netherlands
  2. 2Department of Public Health, Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands
  3. 3Public Health Laboratory, Public Health Service of Amsterdam, Amsterdam, The Netherlands
  4. 4Department of Research, Online Research and Prevention Unit, Cluster of Infectious Diseases, Amsterdam Health Service, Amsterdam, The Netherlands
  5. 5Epidemiology & Surveillance Unit, Centre for Infectious Disease Control, National Institute of Public Health and the Environment, Bilthoven, The Netherlands
  1. Correspondence to Dr Hannelore M Götz, Division of Infectious Disease Control, Rotterdam-Rijnmond Public Health Service, PO Box 70032, Rotterdam 3000 LP, The Netherlands; hm.gotz{at}


Objectives Repeated infections of Chlamydia trachomatis may be new infections or persistent infections due to treatment failure or due to unresolved infections in sexual partners. We aimed to establish the value of using high-resolution multilocus sequence typing (CT-MLST) to discriminate repeated C trachomatis infections.

Methods Paired C trachomatis positive samples (baseline (T0) and after 6 months (T1)) were selected from two Dutch screening implementation studies among young heterosexual people. Typing with six CT-MLST loci included the ompA gene. The uniqueness of strains was assessed using 256 reference CT-MLST profiles.

Results In 27 out of 34 paired cases, full sequence types were obtained. A multilocus (13 cases) or single locus variant (4 cases) was seen, indicating 17 new C trachomatis infections at T1. The ompA genovar was identical for 5 of 17 discordant cases. The 10 cases with concordant typing results were categorised as treatment failure (5 cases) versus persistent or recurrent infections (5 cases). Surprisingly, these concordant cases had C trachomatis strains that were either unique or found in small clusters. The median time between T0 and T1 did not differ between the concordant and discordant cases.

Conclusions High-resolution typing was superior in discriminating new infections compared with only using ompA genovar typing. Many cases (37%) showed exactly the same C trachomatis strain after 6 months. CT-MLST is not conclusive in distinguishing recurrent infections from treatment failure.

  • Chlamydia Trachomatis
  • Molecular Typing
  • Epidemiology (Molecular)

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Repeated infections with Chlamydia trachomatis are common in many sexually active people and remain a challenge in the control of C trachomatis.1 ,2 They can be caused by recurrent infections from an untreated sexual partner in an ongoing partnership, by a new infection in the regular partnership, new infections transmitted by a new sexual partner, or by persistent infections and treatment failure (TF).1 ,3 Previous studies on repeated infections used serovar typing based on the ompA gene variation.1 ,4 ,5 In recent studies more detailed typing of C trachomatis was introduced in which other polymorphic genes of C trachomatis were also analysed.6 ,7 A high-resolution typing technique combining sequences of the ompA gene and five other genes was validated and named C trachomatis multi locus sequence typing (CT-MLST).6 ,8 ,9 Using CT-MLST on large sample sizes showed that different C trachomatis strains were found among heterosexual individuals compared with homosexuals whereas these strains had the same ompA genovar type. Therefore using ompA typing was not sufficiently discriminatory.10 ,11

In this study, we opted for CT-MLST to see if this method was useful to study repeated C trachomatis infections. Samples from two previous Dutch studies were used. First, the systematic internet-based Chlamydia Screening Implementation (CSI) programme in the Netherlands, in which all C trachomatis-positive participants automatically received a retest kit after 6 months.12 Second, the Chlamydia Retest Implementation (CRI) study among heterosexual patients with C trachomatis infection at the STI clinic Rotterdam who were invited for a repeat test after 4–5 months.13 The additional value of CT-MLST compared with ompA genovar typing was studied to distinguish recurrent or persistent infections from new infections.


CSI study

The CSI study was a systematic, internet-based screening programme running from March 2008 to February 2011.14 ,15 Sexually active people aged 16–29 years old were invited for annual screening in Rotterdam, Amsterdam and the South-Limburg province. Participants with Chlamydia infection received a letter with the advice to get treatment for themselves and their current partner. They were requested to answer additional questions about treatment and partner notification 10 days after receiving their results. Repeat test kits were automatically sent after 6 months to all previously infected participants. Data collection was at baseline test (T0) and first retest (T1), and included socio-demographic and behavioural variables, especially recent partners (from the previous 6 months).12 A questionnaire at T1 was only introduced in the last CSI screening year, therefore data at T1 were not available for most participants. Informed consent for additional research was asked for each sample collected.

CRI study

In the CRI study at the Rotterdam STI clinic (March–December 2011), heterosexual patients with a C trachomatis infection received treatment and were invited for repeat testing after 4–5 months, either by home-based sampling (mailed test kit) or by sampling at the clinic.13 All participants filled out an internet-based questionnaire to collect demographic and behavioural data at T0 and T1. All provided consent for the use of their samples for research.

Background population of heterosexuals in Amsterdam

A reference group was selected from a previous study that consisted of C trachomatis positive samples from 256 adult male and female heterosexuals visiting the STI clinic in Amsterdam. Samples were collected during 6 months in 2009 and 2010.10

Screening assays and quantification of C trachomatis

C trachomatis DNA testing in Rotterdam was performed by strand displacement amplification (Becton, Dickinson and Company, Breda, The Netherlands). In Amsterdam transcription mediated amplification, Chlamydia single and Combo assays were used (Hologic/GenProbe, San Diego, USA). DNA was extracted from clinical samples using 200 µL manufacturers transport medium, from swabs or urine that had been tested positive for C trachomatis by screening analysis.8 DNA solutions were tested and semi-quantified with the in-house pmpH lymphogranuloma venereum (LGV) quantitative PCR for the presence of genomic C trachomatis DNA.16

MLST and cluster analysis

DNA isolates were amplified by nested PCR, sequenced and analysed for the regions ompA, CT046 (hctB), CT058, CT144, CT172 and CT682 (pbpB), as previously described.8 ,10 Each new locus was given a unique allele number. The multilocus sequence type (CT-MLST-6) was checked against the C trachomatis database ( and numbered accordingly. With this typing assay multiple infections could not be analysed Cluster analysis was performed using BioNumerics 7 (Applied Maths, Sint-Martens-Latem, Belgium) and a minimum spanning tree was generated using the sequence types.

For concordant genotypes, all six loci were identical between the two C trachomatis infections in one person. Discordant genotypes between two C trachomatis positive samples of one person were noted in case of single locus (SLV) or multilocus variants (MLV).

Discordant genotypes at T0 and T1 defined a new infection in a person. Concordant genotypes were assumed to be consistent with recurrent infection from untreated partner, persistent infection or TF.


In 75 CRI participants, 13 repeated infections were found. Coupled samples could be fully genotyped for 11 cases (table 1). In the CSI study, we found 242 repeated infections in 2756 participants. Of these 242 participants, 85 (35%) gave consent for further research at T0 and 40 of those (47%) also gave consent for use of the T1 sample. Samples from 19 of these 40 cases were missing at either T0, T1 or at both time points. We were able to type 16 of the remaining 21 coupled samples (table 1). In the CSI population we did not find a significant difference in sex, age or ethnicity between those giving consent or not (data not shown).

Table 1

Selection of included participants from two studies in the Netherlands

The CT-MLST sequences of the 27 cases from both studies with fully typed pairs were used to generate a minimum spanning tree (figure 1). To determine whether the C trachomatis strains from these 27 participants represent strains commonly circulating in the Netherlands, we added MLST sequences of 256 reference samples.10 The samples with the C trachomatis strains from the 27 cases show up in large clusters, but also in smaller clusters, or as singletons that represent unique C trachomatis strains.

Figure 1

Minimum spanning tree of 27 cases with typing data of Chlamydia trachomatis strains at T0 and T1. Sizes of the node discs are proportional to the number of samples of each sequence type. Branches show one locus difference, halos indicate clusters. OmpA genovar types are indicated. Green: fully concordant strains at T0 and T1; one case number for two samples. Orange: single locus variant between T0 and T1 strain; two case numbers. Red: discordant C trachomatis strains at two to five loci between T0 and T1; two case numbers. White: background population of 256 heterosexual men and women in Amsterdam.10 Grey: a cluster is defined by all strains that are identical or connected by a single locus variant. Genovar (ompA) types are indicated in general per large cluster.

Concordant repeated C trachomatis infections

Of the 27 cases, 10 had fully concordant genotypes at T0 and T1 by MLST typing and consequently the same ompA genovar. These concordant samples are shown in green in figure 1, with one number per case. A unique MLST (singleton) was found for cases 2, 3 and 5. Cases 4 and 8 clustered with only one other sample, whereas cases 1, 6, 7, 9 and 10 were found in relatively small clusters. So for all 10 concordant cases, the C trachomatis strains were not highly prevalent in the reference population.

For most of the concordant cases it was difficult to definitively establish the cause of the repeated infection. Table 2 shows more detail. Five index cases (3, 4, 7, 9 and 10) reported to have been treated for C trachomatis infection at T0, as was their regular partner, indicating TF, since the same strain was found at T1. For cases 3 and 4 the partner data were too limited to be conclusive (table 2). Cases 7 and 10, however, reported to only have had sex with the treated regular partner since T0, but without a condom. Case 9 had had sexual risks with new partners and unprotected sexual contact with the old treated partner and her strain showed up in a larger cluster (figure 1).

Table 2

Chlamydia trachomatis genotypes in relation to epidemiological data

Case 8 reported only the same regular partner (as in T0) at T1, so her infection was probably a recurrent infection by this untreated partner. In the four remaining concordant infections (cases 1, 2, 5 and 6) there was neither information on index or partner treatment nor on the number of partners at T1. As epidemiology was inconclusive for these repeated infections, we conclude that these were either persistent within the index case or recurrent infections between partners.

Discordant repeated C trachomatis infections

Of the 17 cases with discordant pairs of C trachomatis strains there were four cases for whom only one of the six loci differed, defined as SLV strains. Notably all four SLV cases (11–14) had the same ompA genovars at T0 and T1 (figure 1, case numbers in orange). The discordancy concerned deletions in locus CT172 for case 11 and in locus CT046 for cases 12 and 14 in the T1 sample. For case 13 there were 12 single nucleotide changes throughout locus CT144. Case 11 had two unique sequence types by MLST, and had risk from the old and a new partner. At least one of the samples from cases 12, 13 and 14 coincided with other samples. Case 13 had only had risk from new partners, confirming the new infection, whereas for cases 12 and 14 epidemiological information was lacking.

For the other 13 of 17 discordant cases, at least two but up to all six loci from the MLST differed between T0 and T1 (figure 1, case numbers in red). It is obvious that these 13 repeated infections were new C trachomatis infections resulting in 26 samples. There were eight samples, representing seven cases (15, 16 both samples, 17, 21, 22, 23, 25) in which a unique C trachomatis strain (singleton) was detected. For five samples (cases 15, 17, 20, 21, 24) the strain clustered with the large genovar E type (MLST 56a). For another five samples (cases 18, 20, 22, 23, 24) the strains clustered with the large genovar F type (MLST 12d). Eight samples were found in smaller clusters (figure 1). Thus, in total there were 10 out of 17 cases that were either at T0 or T1 infected with highly prevalent strains.

In 4 of 13 MLV cases (16, 17, 18, 21) there was no information on treatment at either T0 or T1 (table 2). While untreated, case 25 reported a negative test in between T0 and T1, and only sex with the regular partner whose treatment status was unknown. It is likely that at T1 she had a new infection by the regular partner who got (re)infected by an unknown source. For case 16, MLST typing proved a new infection in contrast to ompA typing. In the eight treated cases, known information about new partners was mostly consistent with finding a new infection. Case 20, however, stated to have had only one partner that was still the regular sex partner at T1. Since a new C trachomatis strain was involved, this probably indicates that the index was reinfected by the partner who had had another unknown infected partner since T0.

Interval time and number of partners

The mean interval between T0 and T1 for all 27 cases was 196 days. For the 10 concordant cases the mean interval was 214 days (95% CI 173 to 256 days); for the 17 discordant cases it was 185 days (95% CI 174 to 196 days).

Notably, most cases with multilocus discordant strains reported to have had a higher number of recent partners at T0. The mean number of partners was 5.4 (95% CI 2.3 to 7.3) whereas this was 1.9 (95% CI 0.8 to 2.9) in the cases with concordant C trachomatis strains. The four cases with SLV strains had a mean number of 3.0 partners (95% CI 0.7 to 5.2). Although we do not know the number of recent partners at T1 for all discordant infections, we note that seven out of eight treated index patients reported two or more partners at T0.


With this extended case study we were able to use for the first time high-resolution sequence typing (CT-MLST) on paired C trachomatis positive samples from 27 people with a repeat infection after a retest offered at 6 months. Including the SLV cases we found that 17 of 27 cases (63%) were newly infected. In 10 cases (37%) the same C trachomatis strain was detected at T1. Interestingly, the C trachomatis strains of almost all concordant cases were found either as a unique genotype or in small clusters.

The CT-MLST also includes genotype determination by the ompA gene and the majority of all samples were of ompA genovars E, D and F, in accordance with previous studies among heterosexuals.10 ,11 Five of the 17 cases with the same ompA type were more accurately discriminated using CT-MLST. This concerned four SLV (cases 11–14) and one multi locus variant (case 16), which signified acquisition of a new infection. Case 16 with genovar E would have been regarded as a concordant infection by using only ompA typing.

Although an SLV was allowed in previous studies for C trachomatis strains to belong to the same strain cluster,8 we opted in this study for the more stringent definition of dealing with a new C trachomatis strain at T1. We chose to classify SLVs as new infections because previous research showed that within one person the genetic stability of the Chlamydia strain is very high.17 Within participants the CT-MLST region’s stability may be comparable to that in culture experiments, as Labiran et al18 showed that these regions remained stable after multiple rounds of cell division in tissue culture. Also a study of sexual couples who transmitted 100% identical Chlamydia strains showed this stability.8 The mutations that we noted for the SLV cases in the present study were rather extensive, either insertion/deletions or many point mutations. We are aware, however, that possibly in these SLV cases the region diverged within the index case (persistent infection) or when transmitted from one partner to the other (recurrent infection with the same Chlamydia strain).

Finding a repeated infection with the same CT-MLST type could simply reflect that a certain strain belonged to a common circulating type. Therefore, a rather unexpected finding was that the 10 types of concordant strains circulated mainly in small clusters or were unique so not highly prevalent strains. However, this finding is in line with the small number of recent partnerships in those with concordant infection at T0.

As an untreated persistent infection and a TF or a recurrent infection from an untreated partner are caused by the same strain it is questionable whether full genome sequencing with maximum sequence discrimination will be conclusive without epidemiological data.17 ,19 It is known that C trachomatis has a rather small and conserved genome, so a typical strain could persist in a population for at least a year without divergence.20 For case four in this study we noted 357 days between T0 and T1 with an identical strain.

In concordant cases for which no information on treatment was available, the repeated infection may be due to recurrent infection by an untreated partner or to persistent infection. In five of six concordant index cases with treatment, partner treatment was also reported and irrespective of the number of partners at T1, this indicated TF. Although we cannot quantify this finding in this study, we hypothesise that TF may occur more frequently than the previously reported TF rate of 8–12% with azithromycin.1 ,3 ,21 ,22

There are some limitations of this study. Although the number of repeated infections in the large-scale CSI project was high (n=242), only very few samples were available for MLST typing due to the procedure that required consent for each sample. A further complication was that not all samples were retrievable after completion of the screening studies. It would be advisable in future studies to ask consent only once and include the possibility of repeated testing.

Our data therefore do not allow extrapolation of the findings to the wider population at the STI clinic or in the screening project and can only be presented as an extended case study. Although genotyping data appear to be missing at random, limiting the potential for bias, we cannot rule out this possibility. As even CT-MLST is not conclusive in all cases, the combination with epidemiological data could provide more certainty in classifying repeated infections of the same genotype. However, the epidemiological data are sensitive to selection bias and reporting bias, as we discussed previously.12 Despite these limitations, our finding, that the majority of repeated infections were new infections, is in concordance with previous studies using ompA typing.1 ,4 ,23 ,24

The finding that samples from 7 of the 34 eligible cases could not be fully typed may have been due to a lower C trachomatis load. It is difficult to obtain 100% fully typed samples since the screening assays are at least 10-fold more sensitive than the typing PCR and sequencing reactions.8 ,16 The screening assays aim at high copy targets like rRNA and cryptic plasmid. The typing assay is directed to low copy chromosome genes. In this study, we found that the cycle threshold values in the quantitative PCR were higher at T1 versus T0 in 21/34 cases (62%), which also indicates that the majority of the repeated C trachomatis positive samples may have had a lower bacterial load after 6 months compared with at baseline. This is an interesting observation found earlier by Walker et al24 as the infections with higher load are expected to contribute more to ongoing transmission. A recent study showed that repeated infections occurred more frequently in those with persisting infection at the time of treatment than in those with spontaneous resolution, presenting evidence of immunity in a subset of people with infection.25 ,26 However, whether C trachomatis genotype differences could influence risk for Chlamydia persistence and/or repeat infections was not studied.

In conclusion, we showed that the more detailed typing method, CT-MLST, was able to resolve more accurately whether people acquired a new C trachomatis infection than using only ompA typing. To ascertain whether the repeated infection is persistent, recurrent by the current partner or TF, when finding exactly the same strain depends on the prevalence of the strain in the population in combination with data on the number of recent partners and treatment. These typing data have implications for directing C trachomatis control strategies, such as risk reduction counselling and improving partner treatment, which all remain essential in preventing repeated infections.

Key messages

  • High-resolution multilocus sequence typing (MLST) enables more accurate distinction between recurrent or persistent infection and new infection with Chlamydia trachomatis than ompA typing.

  • Long-lasting infection with C trachomatis either by treatment failure or continued reinfection occurs frequently (up to 37%).

  • CT-MLST is not conclusive in distinguishing recurrent infections from treatment failure; prevalence of the concordant strain in the population and epidemiological data are needed.


The CSI study project group: Jan van Bergen, Christian Hoebe, Eline Op de Coul, Sander van Ravesteijn, Rik Koekenbier, Elfi Brouwers). Michelle Himschoot (Public Health Laboratory, Amsterdam). Norbert Vaessen and Ad Luijendijk for sample handling at the Erasmus MC in Rotterdam.


Supplementary materials

  • Abstract in Dutch

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Jackie A Cassell

  • Contributors HMG was the author responsible for the design of the reinfection study within CSI and the CRI study. HF contributed to the design. HMG and IvdB analysed CSI data; HMG and MEGW analysed CRI data. RJMB did the MLST genotyping and cluster analysis under supervision of SMB. AGCLS was involved in testing of samples in Amsterdam and contributed to analysis of MLST data. HMG and SBB drafted the paper. All authors commented on draft versions and approved the final version.

  • Funding The genotyping work was supported by ZonMW, Ministry of Health, The Netherlands, grant number 124000001.

  • Competing interests None.

  • Ethics approval Both studies were approved by Medical Ethical Committees (Erasmus University of Rotterdam CRI, Free University Amsterdam CSI).

  • Provenance and peer review Not commissioned; externally peer reviewed.