Testing for Sexually Transmitted Infection using Wet and Dry Self-Collected Brush Samples among Women in Mombasa, Kenya

Abstract

We compared detection of Chlamydia trachomatis (CT), Neisseria gonorrhoeae (GC), and Trichomonas vaginalis (TV) using dry and wet self-collected samples using brushes among females who engage in sex work in Mombasa, Kenya. Detection of TV and GC in dry and wet samples was similar, but CT detection in dry samples appeared lower.

Short summary:

Self-collected samples have potential to increase access to STI testing in low-resource settings, but further validation is needed to determine if dry performs comparably to wet self-collection for STI detection.

INTRODUCTION

Globally, over 350 million cases of curable sexually transmitted infections (STIs) occur annually, of which the most common include Trichomonas vaginalis (TV), Chlamydia trachomatis (CT), and Neisseria gonorrhoeae (GC).¹ STI testing in low-resource areas is low due to structural and other barriers (e.g., costs, stigma, embarrassment, geographical accessibility).^2,3 Simple and scalable interventions to improve STI testing coverage in low-resource settings are needed.

Self-collection is an innovative way to improve coverage of STI testing, as an accessible and cost-effective screening option compared to provider-based collection through traditional pelvic exams.⁴ Most studies on self-collected sampling for STI testing have been conducted in high-income countries, although several studies have demonstrated high feasibility and acceptability of self-collection methods in sub-Saharan Africa^3,5. Our previous study in Nairobi, Kenya found strong agreement between self- and provider-collected samples, both immediately stored in preservation media, for the detection of multiple STIs including CT, GC, and TV in a population of 350 females who engage in sex work (FSWs).²

Self-collected samples for STI testing commonly use a preservation media (sc-WET) to maintain sample integrity. Alternatively, self-collected samples may be collected without the use of preservation media (sc-DRY), which may simplify collection for patients and alleviate concerns about sample spillage.^6,7 There is limited research comparing agreement between these two self-collection methods (sc-WET and sc-DRY) for the detection of TV, GC, and CT in low-resourced settings. To address this gap, we present results comparing the performance of Aptima STI testing (Hologic Inc., San Diego, CA) using sc-DRY and sc-WET samples collected using cytobrushes among FSWs in Mombasa, Kenya as part of an HPV testing parent study. We additionally include comparisons with provider-collected samples.

MATERIALS AND METHODS

Study population

From August 2013 to April 2018, FSWs participating in a cohort study of women at high risk of acquiring STIs and HIV were invited to participate in this cross-sectional study in Mombasa, Kenya. Clinical procedures, including self- and provider-collection of genital samples for STI testing were performed at the Ganjoni Clinic in Mombasa,⁸ which has been the primary research site for the Mombasa Cohort⁹ and venue for STI testing and treatment among at-risk and HIV-positive women in Mombasa for over 25 years.

Study inclusion criteria into the Mombasa Cohort included: i. women aged 18 years and above; ii. residing in the Mombasa area; iii. self-identifying as exchanging sex for payment in cash or in-kind at the time of enrollment; and iv. able to provide informed consent. We conducted enrollment stratified by HIV status to ensure half of the participants were HIV-positive. Exclusion criteria were current pregnancy or a history of hysterectomy or treatment for cervical precancer. All women in the Mombasa cohort approached and screened for this STI study agreed to participate. Once the recruitment goal of 400 women was reached, recruitment was stopped. This study was conducted in a screening population, and women were tested regardless of the presence of symptoms. The study, part of an HPV testing parent study, was approved by the ethical review committees of the University of Nairobi/Kenyatta National Hospital, the University of Washington, and the University of North Carolina. All participants provided written informed consent before participation.

Sample collection and laboratory testing

To perform self-collection of cervico-vaginal samples, a study nurse provided verbal instructions and pictorial diagrams with detailed instructions for self-collection in a private room (Appendix 1). Each woman performed self-collection using two different sample brushes: (1) the Evalyn cytobrush¹⁰ (Rovers, Netherlands) for sc-DRY and (2) the Viba cytobrush¹¹ (Rovers, Netherlands) for sc-WET, which included a plastic cryovial containing 1 ml of Aptima specimen transport media (STM, Hologic), into which swabs were placed immediately by participants after collection. To minimize potential bias from the order of sample collection, women assigned odd study numbers used the Evalyn cytobrush first of the two self-collections, while those with even numbers used the Viba cytobrush first.

After self-collection, a clinician performed a speculum-assisted pelvic examination using a collection brush¹² (Hologic) to collect cervical samples. Similar to the Viba brush (for sc-WET), provider-collected samples were immediately stored in Aptima STM prior to laboratory testing.

All samples were transported daily by project personnel in an 8° C cooler box to the laboratory at the Coast Provincial General Hospital (CPGH) for testing. Upon arrival at CPGH, all samples were transferred into Aptima STM within 12 hours post-collection and stored in a freezer at −80 °C until laboratory testing. Provider- and self-collected samples were tested for CT and GC RNA using the Aptima Combo 2 assay (Hologic); and for TV RNA, the Aptima TV assay (Hologic).

Statistical analyses

Agreement between STI positivity in sc-WET samples compared to sc-DRY and compared to provider-collection was measured using percent agreement and the κ (Kappa) statistic with 95% confidence intervals. Pairwise comparisons using McNemar’s test were conducted to assess differences in STI prevalence in sc-DRY and sc-WET sample samples and in sc-WET versus provider-collection.

RESULTS

Of 400 participating FSWs, 1 woman was missing STI testing results, resulting in a final sample of 399. Overall, the median age of the participating 399 FSWs was 39 years (range 19–66) and 48% (n=193) were HIV-positive. The median frequency of reported vaginal intercourse in the past week was 2 (range 0–60), and median frequency of use of a condom with vaginal intercourse in the past week was 1 (range 0–56).

TV prevalence based on wet and dry self-collected samples was 6.0% and 4.8%, respectively (Figure 1). TV prevalence based on provider-collected samples was 4.3%. CT prevalence was 3.5% in sc-WET, 0.5% in sc-DRY, and 2.8% in provider-collected samples. GC prevalence across sampling method was largely consistent: sc-WET: 2.0%, sc-DRY: 2.8%, and provider-collected: 2.5%. We compared detection between women who were assigned sc-DRY and sc-WET as the first or second self-collection and found no differences in detection based on sample collection order.

Figure 1:

Prevalence of STI by collection method (n=399)

When evaluating concordance of sc-WET with sc-DRY samples, we observed high agreement for TV (K:0.83, 95% CI: 0.70–0.95) and GC testing (K: 0.73, 95% CI: 0.50–0.96) (Table 1). For CT, about 4% (14/399) of women tested positive using sc-WET, compared with less than 1% (2/399) using sc-DRY (p=0.002); concordance of sample types was low (K:0.12, 95% CI: −0.10–0.34). Self-collected samples stored wet and provider-collected samples had moderate concordance when testing for GC (K:0.55, 95% CI: 0.27–0.83) and TV (K: 0.66, 95% CI: 0.50–0.84). However, the concordance for CT was relatively low (K:0.38, 95% CI: 0.13–0.63), although the prevalence of CT was not significantly different across sc-WET and provider-collected sample types (McNemar’s p: 0.61). Concordance of sc-DRY and provider-collected samples testing was moderate for TV (K:0.59, 95% CI: 0.40–0.79) and GC (K: 0.56, 95% CI: 0.30–0.82).

Table 1:

Sexually Transmitted Infection detection agreement of STI samples (n = 399)

Self-collected samples stored wet (SC-WET) versus stored dry (SC-DRY)
	Total		Total	sc-WET +		sc-WET +		sc-WET −			sc-WET −		Kappa (95% CI)	P *
	sc-WET+		sc-DRY+	sc-DRY +		sc-DRY −		sc-DRY +			sc-DRY −
Trichomonas	24		19	18		6		1			374		0.83 (0.70–0.95)	0.13
Chlamydia	14		2	1		13		1			384		0.12 (−0.10–0.34)	0.002
Gonorrhea	8		11	7		1		4			387		0.73 (0.50–0.96)	0.38
Self-collected samples stored wet (SC-WET) vs. provider-collected (PC) samples
		Total	Total		sc-WET +		sc-WET +		sc-WET −	sc-WET −		Kappa (95% CI)		P *
		sc-WET+	PC +		PC +		PC −		PC +	PC −
Trichomonas		24	17		14		10		3	372		0.66 (0.50–0.84)		0.09
Chlamydia		14	11		5		9		6	379		0.38 (0.13–0.63)		0.61
Gonorrhea		8	10		5		3		5	386		0.55 (0.27–0.83)		0.73
Self-collected samples stored dry (SC-DRY) vs. provider-collected (PC) samples
		Total	Total		sc-DRY +		sc-DRY +		sc-DRY −	sc-DRY −		Kappa (95% CI)		P *
		sc-DRY+	PC +		PC +		PC −		PC +	PC −
Trichomonas		19	17		11		8		6	374		0.59 (0.40–0.79)		0.59
Chlamydia		2	11		1		1		10	387		0.15 (−0.12–0.41)		0.01
Gonorrhea		11	10		6		5		4	384		0.56 (0.30–0.82)		0.73

^*P-value based on McNemar’s test, sc=self-collection

DISCUSSION

This study is unique as it directly compares agreement between self-collected samples stored wet and stored dry for transcription mediated amplification (TMA) testing of GC, CT, and TV in sub-Saharan African women. Self-collected samples stored dry for the detection of TV and GC had high concordance when compared with those collected and placed into preservation media. However, we found that sc-DRY methods for the detection of CT had a relatively low concordance with both sc-WET and provider-collected methods, despite careful handling of dry swabs and transfer into media within 12 hours post-collection.

Our study adds to the literature by comparing performance of sc-DRY and sc-WET collection methods in clinical samples for STI detection with Aptima testing; dry and wet swabs for STI detection using TMA have been previously compared using simulated swabs.¹³ That study found that artificially-infected dry swabs exposed to high temperatures had no effect on CT RNA detection using two Aptima assays.¹³ In a US study using clinical samples, dry self-collection methods for the detection of GC and CT DNA using PCR testing was also found to be comparable to sc-WET methods.¹⁴ These results are similar to ours for GC, but CT detection in our study was notably lower for sc-DRY compared to both sc-WET and provider-collected samples. This difference may be explained by the lower copy number of rRNA for CT, the TMA target, and sample laboratory treatment may need to be more robust to detect smaller copy numbers. Other differences in bacteria may also lead to degradation of CT RNA compared with GC and TV. CT titers can also be low, especially in asymptomatic women, making detection more difficult. For HPV testing, several studies in various settings have found that dry collection methods are a valid alternative to wet collection methods.^7,8 In our previous work in this cohort, we found that sc-DRY performed similarly to sc-WET for the detection of high-grade squamous intraepithelial lesions or more severe (≥HSIL) for HPV testing.⁸

Our finding of higher yield in self-collected samples is consistent with our previous work with FSWs in Nairobi, Kenya.² In screening studies such as this one, it may be acceptable that the self-collected samples have higher sensitivity and lower specificity for STI detection. For example, in low-resource settings where women may be infrequently screened, a false positive from a self-collected sample that is later not confirmed is preferable to a false negative that places patients and their sexual partners at risk for disease.

Among potential study limitations, our study had a relatively low prevalence of STIs, particularly CT and TV, compared to other studies of FSW.^2,15–17 Our previous work with FSWs in Nairobi found STI prevalence of 5.2% for CT and 9.2% for TV,² and another study with the Mombasa cohort (among HIV negative FSWs) found a prevalence of 5.5% for CT.¹⁷ Wenote that while the brush heads were identical for the two self-collection methods, Evalyn for sc-DRY and Viba for sc-WET, the collection devices themselves were not (the Viba brush was designed to use a transport medium),⁸ which could have influenced patient’s ability to collect a sample and thus affect test performance. We also note that a different brush was used for provider-collected samples¹² than for the self-collected samples. However, STI detection using self-collected and provider-collected brush samples has been found to be similar,¹⁸ and the validated Aptima tests used in this study are highly sensitive, likely overcoming any potential differences in efficiency of collection devices.^19,20 And although provider-collection was always conducted after self-collection, we believe it unlikely that collection order impacted detection, given the samples were taken from two different anatomical locations and high Aptima test sensitivities.² In some settings, getting a dry sample into transfer media within 12 hours may not be feasible. Lastly, we do not have data available to compare titer values or quantify detection in our samples, so we cannot assess whether low titers may have affected detection.

In conclusion, good concordance of dry and wet self-collected samples was observed for TV and GC detection, in contrast to CT detection, which had relatively low concordance. While we found that sc-DRY performed poorly for CT, prior studies using simulated swabs with Aptima testing, and clinical swabs with PCR testing, found that dry self-collection performed as well as wet for the detection of CT. Additional clinical studies are needed to confirm comparability between dry versus wet self-collected methods for the detection of STIs using Aptima testing; however, the reduced performance of dry self-collection for CT detection may indicate that wet self-collection outweighs any benefits of the delayed transfer to sample media that dry collection allows. Comparability between collection methods may be enhanced in a sample with a larger number of women testing positive for STIs to improve reliability of estimates. Self-collection methods for STI testing can increase access to testing in low-resource settings and high-risk populations, but further research using clinical samples is needed to compare sc-WET and sc-DRY for detection of STI RNA, especially in low-resource settings.