ABSTRACT
Screening for Chlamydia trachomatis and Neisseria gonorrhoeae at the pharyngeal, urogenital, and anorectal sites is recommended for men who have sex with men (MSM). Combining the three individual-site samples into a single pooled sample could result in significant cost savings, provided there is no significant sensitivity reduction. The aim of this study was to examine the sensitivity of pooled samples for detecting chlamydia and gonorrhea in asymptomatic MSM using a nucleic acid amplification test. Asymptomatic MSM who tested positive for chlamydia or gonorrhoea were invited to participate. Paired samples were obtained from participants prior to administration of treatment. To form the pooled sample, the anorectal swab was agitated in the urine specimen transport tube and then discarded. The pharyngeal swab and 2 ml of urine sample were then added to the tube. The difference in sensitivity between testing of pooled samples and individual-site testing was calculated against an expanded gold standard, where an individual is considered positive if either pooled-sample or individual-site testing returns a positive result. All samples were tested using the Aptima Combo 2 assay. A total of 162 MSM were enrolled in the study. Sensitivities of pooled-sample testing were 86% (94/109; 95% confidence interval [CI], 79 to 92%]) for chlamydia and 91% (73/80; 95% CI, 83 to 96%) for gonorrhea. The sensitivity reduction was significant for chlamydia (P = 0.02) but not for gonorrhea (P = 0.34). Pooling caused 22 infections (15 chlamydia and 7 gonorrhoea) to be missed, and the majority were single-site infections (19/22). Pooling urogenital and extragenital samples from asymptomatic MSM reduced the sensitivity of detection by approximately 10% for chlamydia but not for gonorrhea.
INTRODUCTION
Chlamydia trachomatis and Neisseria gonorrhoeae are two of the most prevalent agents of bacterial sexually transmitted infections globally (1). They are associated with urethritis, proctitis, and epididymitis in men, and they also facilitate the transmission of human immunodeficiency virus (2). Both infections are increasing in high-income countries such as the United States (3) and the United Kingdom (4) and in Australia (5), particularly among men who have sex with men (MSM). In 2019, the Australian guidelines recommended three-monthly screening for C. trachomatis and N. gonorrhoeae at all three anatomical sites (pharyngeal, urogenital, and anorectal) among MSM regardless of sexual risk (6). The majority of extragenital C. trachomatis and N. gonorrhoeae infections found in MSM are asymptomatic (7), and regular screening is necessary to identify and treat these infections to prevent ongoing transmission. Past studies have found that urogenital-only testing of MSM misses more than 50% (62 to 84%) of infections (8–10). However, testing three anatomical sites each time results in a large cost for health services (11).
Nucleic acid amplification tests (NAATs) are currently the gold standard method for detecting C. trachomatis and N. gonorrhoeae, as they have high sensitivities and specificities in both urogenital and extragenital samples (12–15). The use of NAATs has resulted in substantially increased numbers of C. trachomatis and N. gonorrhoeae diagnoses (16); however, the costs of testing have also increased with their use, as they are more expensive than culture (17). Pooling samples is a potential strategy that may reduce these costs significantly, as two or more samples from the same individual can be combined into a single sample which is then tested.
Several large pooling studies have been done over the last 2 decades (18–22); however, most focused on pooling urine samples from multiple individuals for large-scale C. trachomatis and N. gonorrhoeae screening in resource-limited settings and settings with low prevalence of C. trachomatis or N. gonorrhoeae. Few pooling studies have included both urogenital and extragenital samples (22–25), and there is still no clear consensus on whether pooling has sufficient sensitivity to be used as a detection tool in at-risk populations, including MSM. A 2018 study in the United Kingdom demonstrated that although cost-effectiveness is a major driver for services to implement pooled-sample testing, the majority of clinicians regard the existing evidence as insufficient to make a decision (26). In this study, we aimed to expand the evidence on the sensitivity of pooled-sample testing for C. trachomatis and N. gonorrhoeae against an expanded gold standard in asymptomatic MSM.
MATERIALS AND METHODS
This prospective study took place at Melbourne Sexual Health Centre (MSHC) between 30 March 2016 and 10 July 2017. MSHC is the largest public sexual health center in Australia, providing more than 40,000 consultations per year, and approximately one-third of the patients are MSM (27). As standard of care at MSHC, all MSM presenting for asymptomatic screening were offered triple-site testing for C. trachomatis and N. gonorrhoeae using a self-collected anal swab, a clinician-collected pharyngeal swab and a urine sample. Each sample was individually tested for C. trachomatis and N. gonorrhoeae using Aptima Combo 2 (AC2) (Hologic Inc., San Diego, CA), which is a duplex NAAT that utilizes transcription-mediated amplification (TMA) to detect a target within the 23S rRNA operon of N. gonorrhoeae and the 16S rRNA of C. trachomatis (14). As standard of care, MSM with a positive result for C. trachomatis or N. gonorrhoeae were recalled to the clinic for treatment. Eligible MSM were approached by one of three experienced research study nurses assigned to the study and were invited to participate. MSM were eligible if they were 18 years old or older, returned to the clinic within 3 weeks of recall, had at least one positive result on study swabs on the day of enrollment, and had not received any antibiotics in the preceding 4 weeks. The median time between the screening and return to the clinic for treatment and providing the study samples was 5 days (interquartile range [IQR], 4 to 6).
After written consent was provided, study nurses collected paired study swabs as described below. All participants were offered treatment on the day after the study swabs were taken.
Sample collection.Each participant agreed to provide two anorectal and two pharyngeal swabs collected by the research nurses and a first-pass urine (FPU) sample collected in a sterile jar by the participant. Research nurses instructed the participants on how to collect a sterile FPU sample. Pharyngeal swabs were taken by swabbing the tonsils or tonsillar crypts and the posterior pharynx. Anorectal swabs were taken by inserting a cotton swab 3 cm into the anus and gently rotating for 5 to 10 s to sample exudates inside the anal ring. For collection of FPU, participants were instructed to provide 20 to 30 ml of the initial urine stream or half a container in a sterile urine collection jar. The minimum amount accepted was 2 ml. One anorectal swab, one pharyngeal swab, and half of a FPU sample from each participant were tested separately to form the reference standard for individual-site testing.
To form the pooled sample, (i) one anorectal swab was placed in an Aptima urine specimen transport tube, vigorously agitated in the buffer for 15 s, rolled firmly on the side of the tube to expel fluid, and discarded, and (ii) one pharyngeal swab was added to this tube, the shaft was broken, and the tube was sealed with the swab left inside. Finally, with a pipette, 2 ml of FPU was transferred to the urine specimen transport tube. Swabs were placed in Aptima buffer on collection, and urine was placed in Aptima buffer within 24 h, as per the manufacturer’s instructions. All samples were kept at room temperature (according to the manufacturer’s instructions) and were tested within 24 h at the on-site laboratory (AC2 assay; Hologic; Panther platform).
Research nurses who collected the paired samples ensured that each swab spent a similar amount of time in contact with the anatomical site of infection. As the first swab may remove bacteria, reducing the bacterial load available for the second swab, participants were assigned a unique study identification number starting from number 1. Odd-numbered participants had the first collected swabs used in the pooled sample, and the even-numbered participants had the second of each swab used in the pooled sample. Sensitivity difference between the odd- and even-numbered participants’ pooled samples was assessed using a chi-square test.
Interpretation of test results.Any sample returning a negative result in the initial AC2 test was considered negative. Any positive results in the initial AC2 were confirmed with an additional TMA-based test: Aptima GC (AGC) for gonorrhea or Aptima C. trachomatis (ACT) for chlamydia. A positive result was defined as a sample testing positive in the initial AC2 as well as on the confirmatory AGC or ACT assay. Samples that were positive in the initial AC2 test but negative or equivocal in the confirmatory AGC or ACT test were reported as indeterminate. Indeterminate or invalid results were considered negative for the purposes of sensitivity analysis.
Statistical methods.We calculated that 100 men with N. gonorrhoeae and 100 with C. trachomatis would provide 80% power (P = 0.05, one sided [paired]) to detect a difference in sensitivity of 10% or less with pooled-sample testing compared to individual-site testing. We assumed that if 90% of recalled MSM still tested positive in the second set of swabs, approximately 111 men with C. trachomatis and 111 with N. gonorrhoeae would need to be recruited.
Sensitivity (and associated 95% confidence intervals [CI]) of pooled samples for C. trachomatis and N. gonorrhoeae was calculated against an expanded gold standard where a positive result in either the pooled-sample test or individual-site test was assumed to represent a true infection. The McNemar test for matched pairs was used to test the statistical significance of the sensitivity difference between pooled-sample and individual-site testing. Stata version 12 was used for statistical analysis (StataCorp, College Station, TX). Ethics approval was obtained from Alfred Hospital Ethics Committee, Melbourne, Australia (approval no. 137/16).
RESULTS
Initially, 202 men with at least one positive C. trachomatis or N. gonorrhoeae result on self-collected samples agreed to participate in the study (C. trachomatis = 119, N. gonorrhoeae = 106). After a median of 5 days, 162 asymptomatic MSM remained with a positive result on individual-site or pooled-sample testing, and these men were enrolled in the study. The median age of the participants was 29 years (IQR, 25 to 35). There were 109 men with C. trachomatis infection and 80 men with N. gonorrhoeae infection. C. trachomatis was detected in 105/109 (96.3%) men on individual-site testing (8 pharyngeal, 21 urogenital, and 80 anorectal) and 4/109 men (3.7%) on pooled-sample testing. N. gonorrhoeae was detected in 77/80 (96.3%) men on individual-site testing (32 pharynx, 5 urogenital, and 51 anorectal) and 3/80 (3.7%) men on pooled-sample testing. Thirty-two of 162 (19.8%) men had a coinfection with both C. trachomatis and N. gonorrhoeae.
Pooled-sample testing detected 86.2% of C. trachomatis infections (94/109 [95% CI, 78.5 to 91.5]) compared to the expanded gold standard. The sensitivity of pooled-sample testing was significantly lower than that of individual-site testing, which detected 105/109 infections (96.3% [95% CI, 90.9 to 98.6]) (P = 0.02) (Table 1). Pooled-sample testing detected 91.3% of N. gonorrhoeae infections (73/80 [95% CI, 83.0 to 95.07]) compared to the expanded gold standard, and the sensitivity of pooled-sample testing was not significantly lower than that of individual-site testing (96.3%, 77/80 [95% CI, 89.6 to 98.7]) (P = 0.34). Overall, pooling failed to detect a total of 22 infections that were detected by individual-site testing. All C. trachomatis infections missed by pooling were single site (n = 15; 3 urogenital, 2 pharyngeal, and 10 anorectal). N. gonorrhoeae infections that were missed by pooling (n = 7) were both single site (4/7; 2 pharyngeal, 1 anorectal, and 1 urogenital) and multisite (3/7; 2 anorectal-urogenital and 1 anorectal-pharyngeal). All 15 C. trachomatis infections missed by pooled-sample testing were negative, with no invalid or indeterminate pooling results. Of the seven N. gonorrhoeae infections missed by pooled-sample testing, five tested negative, one was invalid, and one was indeterminate. The two pooled samples that showed an invalid and an indeterminate result could not be repeated, and these were considered negative, as indicated in Materials and Methods.
Comparison between pooled-sample and individual-site testing for Chlamydia trachomatis and Neisseria gonorrheae
The proportion of detected infections did not vary by the order of sampling. Pooling detected 89.5% of C. trachomatis infections in odd-numbered participants (51/57 [95% CI, 78.9 to 95.1]) compared to 82.7% in even-numbered participants (43/52 [95% CI, 0.3 to 90.6]) (P = 0.41). Similarly, pooling detected 94.6% of N. gonorrhoeae in odd-numbered participants (35/37 [95% CI, 82.3 to 99.0]) compared to 88.4% in even-numbered participants (38/43 [95% CI, 75.5 to 94.9]) (P = 0.78). The P values for the sensitivity differences were not statistically significant; however, the study was not powered to detect differences of less than 10%.
DISCUSSION
In this prospective study of asymptomatic MSM being tested for C. trachomatis and N. gonorrhoeae across three anatomical sites, we found a 5 to 10% reduction in sensitivity with pooled-sample testing compared to the expanded gold standard. Chlamydia detection was less sensitive when pooled samples were used (86.2% versus 96.3%). For gonorrhea infections, however, we did not find a significant sensitivity loss with the use of pooling (91.3% versus 96.3%).
To our knowledge, there have been four studies assessing pooling of genital and extragenital samples in MSM (22–25). Sultan et al. conducted the largest study of pooling to date in MSM in the United Kingdom using self-collected samples and the AC2 assay (22). We evaluated the same assay with clinician-collected pooled samples. Sultan’s pooled-sample sensitivity for detecting C. trachomatis in asymptomatic MSM was similar to our figure (89% versus 86%, respectively); however, our sensitivity for detecting asymptomatic N. gonorrhoeae was higher (91% versus 82%) (28). Both studies had a high proportion of pharyngeal infections in the N. gonorrhoeae arm, which are known to have lower bacterial loads than infections at other anatomical sites (29). Swabbing technique is important for detecting pharyngeal gonorrhea (28), and it is likely that our clinician-collected pharyngeal samples yielded a higher organism load, which in turn improved the performance of pooled-sample testing.
In addition, we also report lower pooled-sample sensitivities than Speers et al. for both C. trachomatis (86% versus 94%) and N. gonorrhoeae (91% versus 100%). The study by Speers et al. included a mix of asymptomatic and symptomatic patients (n = 107) in an inner-city sexual health clinic in Australia (24). It is possible that our lower numbers were due to our having an asymptomatic cohort while Speers et al. had a mixed cohort. De Baetselier et al. showed no significant sensitivity loss with pooling (25), and Thielemans et al. reported the sensitivity of pooling as 89.5% using a NAAT; however, that study had a small sample size of positive participants (n = 19), and it is difficult to make a comparison, as the authors did not report individual sensitivities for C. trachomatis and N. gonorrhoeae (23).
Although the AC2 assay has high sensitivity for both C. trachomatis and N. gonorrhoeae and is able to detect very small amounts of DNA (0.005 inclusion-forming units/ml for C. trachomatis and 0.10 CFU/ml for N. gonorrhoeae) (30), we observed some missed infections with pooled samples. Low-load infections detected on individual-site swabs may become undetectable in the pooled sample after dilution with urine. We did not test the impact of different urine volumes on the sensitivity of pooling; however, this was not found to be a significant factor in a previous study (22). Previous studies that used NAATs as well as our study demonstrate a sensitivity loss of up to 20%, which suggests that pooling may result in inherently limited sensitivity regardless of the molecular assay used. There were also seven infections detected by pooled-sample testing but missed by individual-site testing. Five of these men had tested positive for rectal C. trachomatis (n = 3) or N. gonorrhoeae (n = 2) in the preceding 2 weeks. We did not measure the bacterial loads in the pooled or individual-site samples; however, it is possible that these infections represented low bacterial loads or transient deposition.
There were several limitations to this study. MSHC is an urban sexual health clinic, and our C. trachomatis and N. gonorrhoeae prevalence is likely to be higher than that in the general population. Although we could not report prevalence figures in this study due to recruiting only patients with a positive test result, it is likely that the cost savings from pooling will partially depend on the prevalence of C. trachomatis and N. gonorrhoeae in the community. The findings from this study may not be generally applicable to clinics with lower C. trachomatis and/or N. gonorrhoeae prevalence. Our findings may also not be applicable to resource-limited settings, as we used a NAAT, which has higher costs. Furthermore, pooling may limit treatment choices in areas where rectal chlamydia is treated with doxycycline instead of azithromycin. A positive result from pooled samples would need to be deconstructed, as the treatment varies by the anatomical site of infection.
We observed that some of the men who were positive in the initial asymptomatic screening tested negative via the study swabs taken after a median of 5 days. This is likely due to overreporting of the initial positives as a result of contamination during self-collection. We excluded patients who had received an antibiotic in the past 4 weeks; thus, the initial positives are unlikely to be overreported due to nonviable organisms detected as a positive signal. Spontaneous clearance of low-load infections is another possibility; however, this is unlikely given the short time between the two episodes of testing.
Overall, we have shown that that using pooled samples in asymptomatic MSM to detect C. trachomatis or N. gonorrhoeae is less sensitive than individual-site testing for C. trachomatis but not for N. gonorrhoeae. Published studies suggest that pooling is potentially promising as an emerging diagnostic method for C. trachomatis and N. gonorrhoeae detection. The main benefit is the significant cost savings that could be achieved using this method at the expense of a small sensitivity loss, particularly in health services with a high proportion of MSM population that require frequent triple-site testing. This is also relevant to our service, where one-third of patients are MSM (27), particularly in the context of the recent changes to the Australian guidelines, where all MSM are now recommended to undergo 3-monthly screening for sexually transmitted infections regardless of sexual risk (6). While pooling studies are done based on the assumption that considerable savings can be achieved by reducing the sample numbers, a comprehensive cost-effectiveness analysis is required to estimate the true extent of savings. Moreover, health services need to consider the cost of untreated infections in the population, given that pooling misses some infections, and untreated asymptomatic infections are an important reservoir of transmission (31).
ACKNOWLEDGMENTS
This study was funded by the Australian National Health and Medical Research Council (NHMRC) Program Grant (GNT568971). E.P.F.C. is supported by an NHMRC Emerging Leadership Investigator Grant (GNT1172873). T.R.H.R. was supported by the NHMRC Early Career Fellowship (GNT1091536).
FOOTNOTES
- Received 2 December 2019.
- Returned for modification 25 December 2019.
- Accepted 29 February 2020.
- Accepted manuscript posted online 4 March 2020.
- Copyright © 2020 American Society for Microbiology.
