Comparison of mRNA-based and next-generation sequencing HPV assays in a population-based cervical cancer screening program in China

Abstract

Objective

To evaluate and compare the diagnostic performance of the Food and Drug Administration-approved APTIMA^® human papillomavirus (HPV) assay (E6/E7 mRNA-based) and the Mygene HPV-STI Assay (MHSA) in detecting cervical intraepithelial neoplasia (CIN) 2+ and CIN3+ lesions among Chinese women in a screening setting.

Methods

A total of 3,166 women were enrolled from Wuxiang County, Shanxi Province, China, and followed for 3 years. APTIMA and MHSA were performed using cervical cytology specimens. Histopathology served as the reference standard. Diagnostic accuracy (sensitivity, specificity, and predictive values), assay concordance, and receiver operating characteristic (ROC) analyses were conducted for baseline and longitudinal detection of CIN2+/CIN3+.

Results

Among 2,683 women with complete follow-up, the prevalence of high-risk-HPV was 17.22% by APTIMA and 20.16% by MHSA. Substantial concordance was observed (κ=0.798). APTIMA showed higher specificity than MHSA when combined with cytology (CIN2+: 88.42% vs. 87.96%, p=0.038; CIN3+: 87.28% vs. 86.87%, p=0.063), while sensitivity and predictive values were comparable. The MHSA demonstrated broader HPV genotyping, while APTIMA yielded slightly superior performance in post-test probability for CIN3+ over 3 years. ROC analysis confirmed comparable diagnostic discrimination.

Conclusion

Both assays offer high clinical accuracy within 3 years. While APTIMA remains the validated standard, the MHSA may serve as a promising next-generation sequencing-based alternative, particularly in expanded genotyping.

Synopsis

A prospective study compared the APTIMA mRNA assay with next-generation sequencing-based Mygene HPV-STI Assay (MHSA) for screening. Both assays showed high sensitivity and agreement; APTIMA had slightly higher specificity. MHSA detected more high-risk human papillomavirus types and pathogens, offering broader genotyping capability.

INTRODUCTION

Cervical cancer ranks as the fourth most common malignancy among women worldwide, with an estimated 661,021 new cases and 348,189 deaths reported in 2022 [1]. In China, cervical cancer remains a major public health concern, with an incidence of 19.3 per 100,000 women and a mortality rate of 12.4 per 100,000 [1]. In alignment with the World Health Organization (WHO) Global Strategy to Eliminate Cervical Cancer, which targets a reduction in incidence to ≤4 per 100,000 women annually [2], China has intensified nationwide screening efforts to accelerate progress toward this goal.

Persistent infection with high-risk human papillomavirus (HR-HPV), particularly types 16 and 18, is the primary cause of cervical cancer. The natural course of human papillomavirus (HPV) infection often involves progression from transient infection to cervical intraepithelial neoplasia (CIN). Without timely intervention, CIN2+ lesions carry a substantial risk of evolving into invasive cervical cancer over years or decades [3]. Therefore, early detection and management of precancerous lesions through screening are essential to disrupt this progression.

Historically, cervical cancer screening methods have included cytology (Pap smear), visual inspection with acetic acid or iodine, and HPV DNA testing. In its 2021 guidelines, the WHO recommends HPV DNA testing as the preferred primary screening method due to its high sensitivity for detecting CIN2+, a robust negative predictive value, and the ability to support risk-based triage strategies [4]. Laboratory-based HPV test results can also inform more refined triage algorithms, thereby improving screening efficiency and resource allocation [5].

Current clinically validated HPV tests for primary screening (combined with cytology or HPV screening alone) include nucleic acid-based assays targeting either viral DNA (e.g., Cobas 4800, BD Onclarity) or oncogene mRNA (e.g., APTIMA^® HPV Assay) [6]. The APTIMA^® Assay (Hologic, Marlborough, MA, USA), which detects E6/E7 mRNA from 14 HR-HPV types, is approved by the U.S. Food and Drug Administration (FDA) for co-testing with cytology in women aged 30–65 years [7]. Notably, E6/E7 mRNA detection offers greater specificity in identifying transforming infections, thereby differentiating persistent from transient HPV infections more effectively than DNA-based methods [8]. International guidelines recognize APTIMA^® for meeting stringent performance standards, including clinical sensitivity and specificity, intralaboratory reproducibility, and interlaboratory agreement, in the primary screening for CIN2+ lesions [6]. However, it cannot differentiate detailed HPV genotypes directly. For instance, HPV types 16 and 18/45 require additional reflex genotyping tests.

Currently, next-generation sequencing (NGS)-based platforms have emerged as promising alternatives, enabling high-throughput, multiplex detection of multiple HPV genotypes [9,10,11]. The Mygene HPV-STI Assay (MHSA), based on targeted NGS, is designed for the simultaneous detection of multiple HR-HPV types and broader pathogen profiling. Despite the proven clinical performance of assays like APTIMA^®, direct comparisons between mRNA-based and NGS-based HPV detection methods in the context of primary screening remain limited.

Therefore, this study aimed to evaluate the diagnostic performance—specifically sensitivity, specificity, and agreement—of the MHSA versus the FDA-approved APTIMA^® HPV Assay, using histopathological confirmation of CIN2+ and CIN3+ as the reference standard. The results are expected to inform the clinical applicability and scalability of NGS-based HPV testing platforms within national cervical cancer screening programs.

MATERIALS AND METHODS

1. Study design

This prospective diagnostic accuracy study was designed with predefined follow-up protocols, where data collection frameworks were established before the performance of index tests (APTIMA^® HPV Assay and MHSA) and the reference standard (colposcopy with histopathological evaluation). Participants were enrolled between June and December 2016, with a 3-year longitudinal follow-up period. Eligible individuals were recruited via cluster sampling from Wuxiang County, Shanxi Province, China, during routine cervical cancer screenings at local healthcare facilities. Consecutive eligible participants were enrolled to constitute the study cohort.

The study was approved by the Ethics Committee of the Chinese Academy of Medical Sciences (NCC-006582), and all participants provided written informed consent.

2. Participants

Inclusion criteria

• Participants were eligible if they met all of the following criteria:

• 1) Aged 21–64 years with an intact uterus;

• 2) In good general health and capable of undergoing routine cervical cancer screening;

• 3) Willing to undergo colposcopy and biopsy;

• 4) Able to understand the study procedures, voluntarily participate, and provide written informed consent.

Exclusion criteria

• Individuals were excluded if they met any of the following conditions:

• 1) Clinically suspected pregnancy or within 8 weeks postpartum;

• 2) History of cervical cancer or pelvic radiotherapy.

3. Test methods

Sample collection and processing

Cervical samples were collected using a broom-type sampler or cell brush/scraper and preserved in PreservCyt solution (ThinPrep^®) using Berberine liquid cytology bottles in accordance with the manufacturer’s instructions.

For the APTIMA^® assay, 1 mL of cytology samples (either pre- or post-ThinPrep 2000 system processing) was transferred into APTIMA sample transfer tubes according to the test kit protocol.

Transport and storage conditions

• 1) Liquid-based cytology samples were transported at 2°C–30°C.

• 2) Samples for APTIMA testing were transferred into sample tubes within 105 days of collection.

• 3) ThinPrep liquid cytology samples were stored at 2°C–30°C, but storage above 8°C was limited to ≤30 days.

• 4) Samples in APTIMA tubes were stored at 2°C–30°C for up to 60 days or at −20°C to −70°C for long-term storage.

• 5) Residual baseline exfoliated cell samples were stored at −80°C for later MHSA testing.

APTIMA^® HPV Assay (Hologic)

Principle: The APTIMA^® assay performs qualitative detection of E6/E7 mRNA from 14 HR-HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68) using target capture, transcription-mediated amplification (TMA), and hybridization protection assay. Internal control (IC)-specific amplicons are detected via “flash” chemiluminescence probe, and HPV-specific signals are identified using time-resolved luminophore probes.

Procedure: 1 mL of PreservCyt-preserved samples was processed according to the manufacturer’s protocol, including ICs for quality assurance.

Positivity threshold: Defined by the manufacturer's cutoff. Detection limits are <100 copies/reaction for most HPV types, including HPV16, 18, 31, 33, 35, 39, 45, 51, 56, 59, and 68. 100–500 copies/reaction for HPV52, 58, and 66. And 20–40 copies/reaction for HPV16, 18, and 45 in the APTIMA^® 16, 18/45 assay.

APTIMA^® 16, 18/45 HPV

Principle: Specifically detects E6/E7 mRNA from HPV types 16, 18/45 using the same TMA-HPA approach.

Procedure: Performed as per the manufacturer's guidelines, incorporating ICs for quality control.

Positivity threshold: Defined by the cutoff values with a detection limit of 20–40 copies per reaction.

MHSA (China)

Principle: Utilizes targeted NGS for detection of 27 HR-HPV types(14 HR-HPV 16, 18, 31,33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68 and 3 probably HR-HPV 26, 53, 82 and 10 LR-HPV 6, 11, 40, 42, 43, 44, 55, 61, 81, 83), and 14 STIs types (Chlamydia trachomatis, Treponema pallidum, Mycoplasma genitalium, Trichomonas vaginalis, Neisseria gonorrhoeae, Herpes simplex virus type 1, Herpes simplex virus type 2, Varicella zoster virus, Mycoplasma hominis, Ureaplasma urealyticum, Ureaplasma parvum, Haemophilus ducreyi, and Chlamydia trachomatis serovars Lymphogranuloma Venereum 1 and Lymphogranuloma Venereum 2).

Procedure: The protocol comprised 4 major steps: DNA extraction; multiplex polymerase chain reaction (PCR) amplification; library construction and purification; and high-throughput sequencing. Residual baseline samples (1 mL) stored at –80°C were subjected to DNA extraction. A volume of 6.3 μL of each extracted DNA was added to a PCR mixture containing primers, DNA polymerase, deoxynucleoside triphosphates, and PCR buffer. A one-step multiplex PCR was then performed in a 15 μL final volume in 96-well plates on a Veriti Thermal Cycler (Thermo Fisher Scientific, Waltham, MA, USA). After amplification, 5 μL of each PCR product was pooled, and 200 μL of the pooled amplicons was purified using DNA Clean Beads (Vazyme, Nanjing, China) according to the manufacturer’s instructions. The purified pool concentration was quantified with a Qubit fluorometer and adjusted according to Illumina library preparation guidelines; library fragment size was assessed by electrophoresis. Finally, the library preparation and sequencing were performed on the KM MiniSeqDx-CN platform using an Illumina Mid-Output sequencing kit (Illumina, San Diego, CA, USA) in accordance with standardized operating procedures [12].

Quality control: 1) Library quality was assessed by concentration and fragment-size analysis; only libraries with DNA concentration ≥0.5 ng/μL and a dominant fragment peak between 380–500 bp were subjected to sequencing. Sequencing runs yielded high-quality data, with ≥75% of bases achieving Q30. Reads were aligned using BWA (version 0.7.17) and filtered to retain only high-confidence alignments (target length ≥100 bp, mismatches <10). 2) Sample validity was confirmed by detection of the IC gene glyceraldehyde 3-phosphate dehydrogenase (≥13 reads). Pathogen-specific positivity thresholds (e.g., HPV16: ≥20 reads) were established by ROC curve analysis. 3) Throughout the process, negative controls (nuclease-free water) and positive controls (reference standards) were included to monitor for contamination and verify assay performance.

Cytology

Cytology results were classified as normal, atypical squamous cells of undetermined significance (ASC-US), low-grade squamous intraepithelial lesion, or high-grade squamous intraepithelial lesion or severe.

Colposcopy and histopathology

Colposcopy referral was indicated for participants with a positive APTIMA^® 16, 18/45 result, baseline cytology ≥ ASC-US, or follow-up cytology ≥ ASC-US. Colposcopy was performed within 12 weeks of cervical cell sample collection. For women with abnormal colposcopy findings, tissue biopsies were obtained. Histopathological results were categorized as cancer, CIN3, CIN2, or less than CIN2.

4. Baseline and follow-up protocols

Baseline: All participants first underwent APTIMA^® HPV testing. For APTIMA^®-positive cases, the APTIMA^® 16, 18/45 assay was subsequently conducted. Cytology was performed for all participants. Colposcopy referral was indicated for those with a positive APTIMA^® 16, 18/45 result or cytology ≥ ASC-US, with histopathological biopsy performed when clinically necessary. Histopathological assessment, considered the reference standard for CIN diagnosis, was conducted by blinded pathologists. CIN2+ was defined as positive and follow-up was terminated at this point, whereas all other outcomes continued to the scheduled follow-up endpoint.

Follow-up: Participants with HR-HPV positivity or cytology ≥ ASC-US were followed annually; Participants negative on both HPV and cytology (co-test negative) were followed once every 3 years. Cytology was repeated during follow-up and the detection of ASC-US or worse prompted repeat colposcopy and biopsy.

At study completion, women who did not progress to ≥ CIN2 were classified as “negative.” Co-test negative cases were considered true negatives due to their low likelihood of harboring significant lesions.

5. Statistical analysis

Descriptive statistics were used to summarize demographic and clinical characteristics: categorical variables were reported as counts and percentages (%); continuous variables with skewed distributions were presented as medians and interquartile ranges (IQRs). To compare HPV positivity rates between APTIMA^® and MHSA (including types 16, 18/45, and overall 14 HR-HPV), McNemar’s chi-square test was used. Agreement between methods was assessed using overall agreement rate, Cohen’s kappa coefficient (κ), and the proportion of positive/negative agreement, each reported with 95% confidence intervals (CI) calculated via asymptotic methods [13]. Diagnostic accuracy measures included sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and referral rates. McNemar’s test was applied to compare sensitivity/specificity; Pearson’s χ² test was used for PPV and NPV comparisons; the Wilson score method was used to compute 95% CIs [14,15]. Poisson regression was used to calculate incidence rate (IR), incidence rate ratio (IRR), and corresponding 95% CIs for CIN2+ within 3 years.

The profile likelihood method was used for CI estimation. Receiver operating characteristic (ROC) curves were constructed to evaluate the diagnostic performance of APTIMA^® and MHSA for detecting CIN2+ and CIN3+.

RESULTS

1. Study population and baseline characteristics

Of the initial cohort comprising 3,283 participants, 3,166 women (96.4%) successfully completed baseline cytology, MHSA, and APTIMA testing and were included in the final analysis. The flowchart of participant enrollment and selection among them was presented in Fig. 1. The 2,683 (84.7%) completed the 3-year follow-up, resulting in an attrition rate of 15.3%. During follow-up, 118 participants were diagnosed with CIN2+. Baseline characteristics are summarized in Table 1. The median age was 45 years (IQR=40–50), and 661 participants (24.63%) were postmenopausal. Approximately 49.31% (1,323/2,683) had a junior high school education, and 99.66% (2,674/2,683) reported no smoking history.

Fig. 1. Flowchart of participant enrollment and selection. All cervical lesions were confirmed by histopathological examination.

ASC-US, atypical squamous cells of undetermined significance; CIN2, cervical intraepithelial neoplasia grade 2; CIN2+, cervical intraepithelial neoplasia grade 2 or worse; HPV, human papillomavirus; LBC, liquid-based cytology; NILM, negative for intraepithelial lesion or malignancy.

Table 1. Baseline demographic characteristics of study participants (n=2,683).

Characteristic		Values
Age of joining the group		45 (40–50)
Menopause		661 (24.63)
Marital status
	Married	2,647 (98.66)
	Widowed	33 (1.23)
	Divorced	3 (0.11)
Educational level
	Never been educated	33 (1.23)
	Primary school	515 (19.19)
	Junior high school	1,323 (49.31)
	High school	343 (12.78)
	University and above	468 (17.44)
	NA	1 (0.04)
Smoking history
	No smoking	2,674 (99.66)
	Smoking now	9 (0.34)
History of drinking
	Never	2,536 (94.52)
	Occasionally	142 (5.29)
	Often	4 (0.15)
	NA	1 (0.04)
3-year cumulative CIN2+		118 (4.40)
3-year cumulative CIN3+		56 (2.09)

Values are presented as median (Q1–Q3) or number (%).

CIN2+, cervical intraepithelial neoplasia grade 2 or worse; CIN3+, cervical intraepithelial neoplasia grade 3 or worse; NA, not available.

2. HPV detection and consistency analysis

The HR-HPV positivity rate was significantly higher with MHSA (20.16%, 541/2,683) than with APTIMA (17.22%, 462/2,683), p<0.001 (Table S1). While MHSA also showed slightly higher positivity for HPV16 (4.29% vs. 4.03%) and HPV18/45 (1.42% vs. 1.12%), these differences were not statistically significant (p=0.169 and p=0.080, respectively). Age-stratified HPV prevalence (Fig. S1) showed peaks in the 50–59-year age group for both assays. Moreover, the distribution of HPV positivity by histopathological diagnosis was similar between APTIMA and MHSA (Fig. S2).

Substantial agreement was observed between the 2 assays for all tested HR-HPV types (Table 2). The kappa coefficient for any 14 HR-HPV types was 0.7980 (95% CI=0.7684–0.8275). For HPV16: κ=0.9111 (95% CI=0.8714–0.9508). For HPV18/45: κ=0.7617 (95% CI=0.6486–0.8749). Overall agreement rates were high: HR-HPV: 93.85% (95% CI=92.88–94.70); HPV16: 99.29% (95% CI=98.90–99.55); HPV18/45: 99.40% (95% CI=99.03–99.63).

Table 2. Agreement between APTIMA and MHSA at baseline.

MHSA result / Agreement measure		APTIMA
		Any 14 HR-HPV positive			HPV 16			HPV 18/45
		Positive	Negative	Total	Positive	Negative	Total	Positive	Negative	Total
MHSA
	Positive	419	122	541	102	13	115	26	12	38
	Negative	43	2,099	2,142	6	2,562	2,568	4	2,641	2,645
	Total	462	2,221	2,683	108	2,575	2,683	30	2,653	2,683
Overall agreement (95% CI)		0.9385 (0.9288–0.9470)			0.9929 (0.9890–0.9955)			0.9940 (0.9903–0.9963)
Kappa (95% CI)		0.7980 (0.7684–0.8275)			0.9111 (0.8714–0.9508)			0.7617 (0.6486–0.8749)
Proportion of positive agreement (95% CI)		0.8355 (0.8210–0.8490)			0.9148 (0.9036–0.9248)			0.7647 (0.7483–0.7804)
Proportion of negative agreement (95% CI)		0.9622 (0.9543–0.9688)			0.9963 (0.9932–0.9980)			0.9970 (0.9941–0.9985)

CI, confidence interval; HPV, human papillomavirus; HR-HPV, high-risk human papillomavirus; MHSA, Mygene HPV-STI Assay.

3. Clinical performance

Detection of CIN2+ and CIN3+ at baseline

At baseline, APTIMA demonstrated higher specificity than MHSA for both CIN2+ (85.08% vs. 81.98%) and CIN3+ (84.07% vs. 81.04%), with statistically significant differences (p<0.001) (Table 3). Although MHSA exhibited numerically lower sensitivity, PPV, and NPV than APTIMA, none of these differences reached statistical significance (p>0.05). Both assays demonstrated higher sensitivity but lower specificity compared to cytology alone. Co-testing (HPV + cytology) achieved the highest combined sensitivity and specificity for detecting CIN2+/CIN3+ lesions. There were no significant differences between MHSA and APTIMA when used with cytology triage (Table 3).

Table 3. Diagnostic performance of HPV and cytology-based strategies for detecting CIN2+ and CIN3+ at baseline.

Pathologic diagnosis	Method	Sensitivity		Specificity		PPV		NPV
		TP/(TP + FN)	% (95% CI)	TN/(TN + FP)	% (95% CI)	TP/(TP + FP)	% (95%CI)	TN/(TN + FN)	% (95% CI)
CIN2+	MHSA 14 HPV	71/75	94.67 (87.07–97.91)	2,138/2,608	81.98 (80.46–83.41)	71/541	13.12 (10.54–16.23)	2,138/2,142	99.81 (99.52–99.93)
	APTIMA 14 HPV	73/75	97.33 (90.70–99.27)	2,219/2,608	85.08 (83.66–86.39)	73/462	15.80 (12.75–19.40)	2,219/2,221	99.90 (99.67–99.97)
	p-value*	0.617		<0.001		0.341		0.983
	Cytology ASC-US+	65/75	86.67 (77.17–92.59)	2,357/2,608	90.38 (89.18–91.45)	65/316	20.57 (16.48–25.37)	2,357/2,367	99.58 (99.22–99.77)
	Co-test-MHSA	74/75	98.67 (92.83–99.76)	2,294/2,608	87.96 (86.66–89.15)	74/388	19.07 (15.47–23.28)	2,294/2,295	99.96 (99.75–99.99)
	Co-test-APTIMA	75/75	100.00 (95.13–100.00)	2,306/2,608	88.42 (87.14–89.59)	75/377	19.89 (16.18–24.22)	2,306/2,306	100.00 (99.83–100.00)
	p-value†	<1.000		0.038		0.884		<1.000
	MHSA with cytology triage	71/75	94.67 (87.07–97.91)	2,426/2,608	93.02 (91.98–93.94)	71/253	28.06 (22.89–33.90)	2,426/2,430	99.84 (99.58–99.94)
	APTIMA with cytology triage	73/75	97.33 (90.79–99.27)	2,436/2,608	93.40 (92.39–94.30)	73/245	29.80 (24.42–35.80)	2,436/2,438	99.92 (99.70–99.98)
	p-value†	0.617		0.165		0.824		0.999
CIN3+	MHSA 14 HPV	40/41	97.56 (87.40–99.57)	2,141/2,642	81.04 (79.50–82.49)	40/541	7.39 (5.48–9.91)	2,141/2,142	99.95 (99.74–99.99)
	APTIMA 14 HPV	41/41	100.00 (91.43–100.00)	2,221/2,642	84.07 (82.62–85.41)	41/462	8.87 (6.61–11.82)	2,221/2,221	100.00 (99.83–100.00)
	p-value*	<1.000		<0.001		0.500		<1.000
	Cytology ASC-US+	40/41	97.56 (87.40–99.57)	2,366/2,642	89.55 (88.33–90.66)	40/316	12.66 (9.44–16.78)	2,366/2,367	99.96 (99.76–99.99)
	Co-test-MHSA	41/41	100.00 (91.43–100.00)	2,295/2,642	86.87 (85.52–88.10)	41/388	10.57 (7.89–14.02)	2,295/2,295	100.00 (99.83–100.00)
	Co-test-APTIMA	41/41	100.00 (91.43–100.00)	2,306/2,642	87.28 (85.96–88.50)	41/377	10.88 (8.12–14.42)	2,306/2,306	100.00 (99.83–100.00)
	p-value†	-		0.063		0.994		<1.000
	MHSA with cytology triage	40/41	97.56 (87.40–99.57)	2,429/2,642	91.94 (90.84–92.92)	40/253	15.81 (11.83–20.81)	2,429/2,430	99.96 (99.77–99.99)
	APTIMA with cytology triage	41/41	100.00 (91.43–100.00)	2,438/2,642	92.28 (91.20–93.24)	41/245	16.73 (12.58–21.91)	2,438/2,438	100.00 (99.84–100.00)
	p-value†	<1.000		0.233		0.907		<1.000

McNemar's χ² test was used to compare sensitivity, specificity. The Pearson χ² test was applied to compare PPV and NPV. The Wilson score method was used to calculate 95% CIs.

For co-testing, results were considered negative only when both HPV and cytology were negative; all other combinations were classified as positive.

For HPV testing with cytology triage, the following rules were applied:14 HR-HPV−: negative; 14 HR-HPV+ and HPV 16/18/45−: positive if cytology ≥ ASC-US; HPV 16/18/45+: positive regardless of cytology.

ASC-US, atypical squamous cells of undetermined significance; ASC-US+, atypical squamous cells of undetermined significance or worse; CI, confidence interval; CIN2+, cervical intraepithelial neoplasia grade 2 or worse; CIN3+, cervical intraepithelial neoplasia grade 3 or worse; FN, false negative; HPV, human papillomavirus; MHSA, Mygene HPV-STI Assay; NPV, negative predictive value; PPV, positive predictive value; TP, true positive.

^*Comparison between MHSA 14 HR-HPV and APTIMA 14 HR-HPV; ^†Comparison between co-testing (MHSA vs. APTIMA) and between HPV with cytology triage (MHSA vs. APTIMA).

Longitudinal detection rates of CIN2 and CIN3+

Longitudinal detection rates are shown in Tables S2 and S3. At baseline, the CIN3+ detection rate was 15.28 per 1,000 women (95% CI=11.25–20.75). During follow-up, CIN3+ detection rates were:1 year: 20.48/1,000 (95% CI=11.34–36.99); 2 years: 8.70/1,000 (95% CI=3.26–23.17); 3 years: 0.38/1,000 (95% CI=0.05–2.73). Concordance for APTIMA+/MHSA+ results was high: Baseline: 88.23% (30/34) for CIN2+, 97.56% (40/41) for CIN3+; 1-year: 86.67% (13/15) and 100% (11/11); 2-year: 60.00% (3/5) and 100% (4/4); 3-year: 50.00% (4/8) and 100% (1/1). Across all time points, APTIMA–/MHSA– concordant results were associated with extremely low detection rates for CIN2+ (≤2.17/1,000) and 0 cases of CIN3+.

IR, IRR of CIN2+ and CIN3+

As shown in Tables S4 and S5, the overall IRs were: CIN2+: 1,466.02 per 100,000 person-years (95% CI=1,217.16–1,746.67); CIN3+: 695.74 per 100,000 person-years (95% CI=529.06–894.20).

APTIMA and MHSA had similar IRRs (p>0.05), while cytology-negative women had significantly higher risk compared to MHSA-negative women: CIN2+: IRR=2.86 (95% CI=1.54–5.71), p=0.002; CIN3+: IRR=10.82 (95% CI=2.13–197.18), p=0.022.

Diagnostic performance via ROC curves

ROC curves assessed the discriminatory ability of APTIMA and MHSA for detecting CIN2+ and CIN3+ both at baseline and over 3 years.

At baseline (Fig. 2), APTIMA and MHSA showed comparable accuracy for CIN2+ detection (AUC=0.94 vs. 0.93, p=0.184 for co-test; 0.95 vs. 0.94, p=0.255 for triage) and for CIN3+ detection (AUC=0.94 vs. 0.93, p=0.041 for co-test; 0.96 vs. 0.95, p=0.257 for triage).

Fig. 2. ROC curves for baseline detection of CIN2+ and CIN3+ using different screening strategies. (A-D) panels illustrate the diagnostic performance of APTIMA and MHSA combined with cytology (co-test or triage) for detecting CIN2+ and CIN3+ at baseline. (A) Performance of the APTIMA and MHSA with co-test cytology for the detection of CIN2+ at baseline; (B) Performance of the APTIMA and MHSA with co-test cytology for the detection of CIN3+ at baseline; (C) Performance of the APTIMA and MHSA with cytology triage for the detection of CIN2+ at baseline; (D) Performance of the APTIMA and MHSA with cytology triage for the detection of CIN3+ at baseline.

CIN2+, cervical intraepithelial neoplasia grade 2 or worse; CIN3+, cervical intraepithelial neoplasia grade 3 or worse; MHSA, Mygene HPV-STI Assay; ROC, receiver operating characteristic.

At 3-year follow-up (Fig. S3), both assays remained similar in performance (CIN2+: 0.87 vs. 0.86, p=0.132 for co-test; 0.88 vs. 0.87, p=0.224 for triage; CIN3+: both 0.89, p=0.041 for co-test; 0.92 vs. 0.91, p=0.238 for triage).

PPP plots

Among the 2,683 participants, the 3-year cumulative prevalence of CIN3+ was 2.09% (56/2,683). Predictive performance across testing strategies for CIN3+ is shown in Fig. S4. Under the co-testing model, the positive post-test probabilities (+PP) of APTIMA and MHSA were 13.53% and 13.14%, respectively. And the negative post-test probabilities (−PP) of APTIMA and MHSA were 0.21% and 0.22%. In the cytology triage testing model, the +PP values of APTIMA and MHSA increased to 20.82% and 19.76%. Additionally, with a PP of 0.21% compared to MHSA's 0.25%.

DISCUSSION

In this prospective diagnostic accuracy study, we evaluated and compared the clinical performance of 2 HPV testing strategies—the mRNA-based APTIMA^® assay and the NGS-based MHSA—for the detection of CIN2+ and CIN3+ in a large population-based cohort with 3-year follow-up, in Wuxiang County, Shanxi Province, China. Our results demonstrate that both assays demonstrated high sensitivity and strong agreement, with APTIMA showing slightly higher specificity, while MHSA provided broader genotyping capacity and additional pathogen detection. ROC analysis further confirmed the comparable AUCs of the 2 assays for both CIN2+ and CIN3+ at baseline and over 3 years.

Current estimates of the HPV infection rates in China vary across geographic regions, ranging from 9.6% to 23.6% [16,17]. In this study, the prevalence of 14 HR-HPV types detected by the APTIMA assay was 17.22% (462/2,683), while that detected by the MHSA was 20.16% (541/2,683). Both rates exceed the prevalence reported in the Shanxi CIN Cohort (8.92%, 900/10,086) [18], which may be attributed to differences in HPV detection methods and population characteristics. Two age peaks in HPV prevalence in women ≤20–25 years and 50–60 years were observed, consistent with previous findings [17,19]. Moreover, HPV positivity rates increased in parallel with the severity of cervical lesions in both the MHSA and APTIMA assay results, corroborating observations from a 4-year surveillance study conducted in Eastern China [20]. The oncogenic potential and associated risk of CIN vary among different HPV genotypes. In China, HPV-16, HPV-52, and HPV-58 are the most prevalent high-risk types [18]. In our study, the respective positivity rates for HPV-52, HPV-16, and HPV-58 were 4.17%, 3.98% and 2.81% (Fig. S5). Compared with the traditional genotyping strategy based solely on HPV-16/18, a risk-stratified approach significantly improved the specificity (99.13% vs. 98.41%, p<0.001) and PPV (26.80% vs. 15.69%, p=0.032) for detecting CIN3+, while reducing the number of colposcopies needed per CIN3+ case (3.73 vs. 6.38) [21]. The ability of MHSA to simultaneously detect 27 HR-HPV genotypes and 14 sexually transmitted pathogens highlights its potential as a comprehensive screening platform.

The APTIMA assay utilizes TMA technology to detect E6/E7 oncogene mRNA from 14 HR-HPV types. It is a validated index test and offers optional genotyping for HPV 16,18/45 as an additional feature [6]. E6/E7 mRNA serves as a direct biomarker of viral oncogenic activity and is closely associated with the progression of cervical lesions. In a formal validation study by Heideman et al. [22], APTIMA fulfilled all international validation criteria, including non-inferior sensitivity and specificity, as well as high reproducibility. Compared with DNA-based assays, mRNA detection more accurately identifies persistent infections with carcinogenic potential [8,23,24]. In this study, the performance of APTIMA aligns with findings from prior international validations, consistently demonstrating high sensitivity and specificity for CIN2+ detection using E6/E7 mRNA assays. Notably, the sensitivity of APTIMA combined with cytology for detecting CIN2+ was higher than that reported in studies conducted in Liaoning Province and Qingdao City (100% [95.1%–100%] vs. 73.4% [65.1%–81.7%]), likely attributed to a satisfactory colposcopy referral rate in our study population [25].

Although MHSA is a newer NGS-based assay, it demonstrated comparable clinical sensitivity and a slightly higher HR-HPV positivity rate. This may reflect the enhanced analytical sensitivity of NGS platforms, which are capable of simultaneously detecting multiple pathogens and genotypes. However, this increased sensitivity may also lead to the detection of transient or clinically irrelevant infections, potentially reducing assay specificity. In this study, the sensitivity and specificity of MHSA for CIN2+ were lower than those reported for the CLART HPV4S assay (also NGS-based), which may be partially explained by differences in study protocols, such as the policy of retesting all invalid samples in the CLART study [26]. Nevertheless, MHSA offers broader detection capabilities than CLART HPV4S, including additional HPV types and other genital pathogens, thereby supporting its utility in more comprehensive screening and co-infection surveillance strategies.

The missed positive case by the MHSA was identified as HPV16-positive by the APTIMA test. MHSA detected only 8 reads for HPV16, below the positivity threshold, and thus reported a negative result. This may be attributed to a low viral load, leading to a false-negative outcome, similar to findings in comparative studies between HC2 and COBAS, where discordant results often occurred near the detection limit [27]. Additionally, since the MHSA was performed on residual specimens, prolonged storage may have affected HPV DNA integrity and viral load.

The findings from this study provide evidence to support the optimization of cervical cancer screening strategies. ROC analysis demonstrated that the clinical performance of both the APTIMA and MHSA was comparable when used in combination with cytology or with cytology-based triage within a 3-year period. Furthermore, HPV testing followed by cytology triage showed lower sensitivity but higher specificity at baseline for detecting CIN2+ and CIN3+, compared with simultaneous HPV and cytology co-testing. These results are consistent with findings from a previous study conducted in Changzhi County, Shanxi Province, China [28]. Over 3 years, the post-test probability of APTIMA combined with cytology triage was superior to co-testing with cytology, suggesting enhanced clinical value for risk stratification over time. However, a key limitation of the APTIMA assay is its inability to provide detailed HPV genotyping directly, as types 16 and 18/45 require reflex testing. This reduces the assay’s utility for personalized risk evaluation.

Notably, when cytology is used as the primary screening method, even a negative result is associated with a higher residual risk of subsequent lesions compared to a negative APTIMA test and MHSA result (IR: 533.33 vs. 186.74 and 120.07, respectively, Table S4). This difference aligns with the inherent characteristics of cytology, which typically has higher specificity but lower sensitivity than HPV testing, and is consistent with current clinical guidelines recommending a shorter follow-up interval for cytology-based screening (3 years) compared to HPV-based screening (5 years).

The MHSA offers advantages in pathogen genotyping and multi-pathogen screening. Its broader detection capabilities and scalability via NGS workflows highlight its promise for integration into precision public health screening strategies. However, it is important to note that this study was conducted in a specific high-risk rural region. Therefore, while MHSA shows promise, its findings require validation in broader, more diverse populations (including urban and multi-provincial cohorts) before its utility can be generalized for nationwide public health planning.

The APTIMA assay is priced at approximately USD 7 per test. In comparison, the estimated consumable cost of MHSA is approximately USD 4–5 per sample with automated nucleic acid extraction [29]. In terms of workflow efficiency, APTIMA enables same-day HPV reporting with a 6.5-hour runtime per 60-sample batch and a daily throughput of up to 300 samples in an extended shift (hands-on time of 2 hours and 55 minutes). In contrast, MHSA processes 384 samples in 29 hours. Within this period, 20 hours are allocated to sequencing, a fully automated process requiring no manual intervention. The remaining 9 hours involve limited hands-on time for sample preparation, amplification, and bioinformatic analysis, ensuring both operational efficiency and reproducibility.

This study possesses several strengths, including a large, population-based cohort, use of histopathology as the reference standard, and a 3-year longitudinal follow-up period, all of which strengthen the robustness of the diagnostic performance evaluation. Moreover, the inclusion of both baseline screening and follow-up outcomes enhances its real-world applicability.

This study has several limitations. First, MHSA was performed on residual specimens, and prolonged storage may have contributed to occasional false negatives. Although long-term frozen storage could theoretically reduce detection sensitivity, our stringent quality control data confirmed acceptable sample quality. Second, the follow-up period was limited to 3 years, while international guidelines recommend 5-year intervals, and the long-term negative predictive value of MHSA requires further investigation. Third, MHSA has not yet undergone international clinical validation, and multicenter studies are needed to confirm its clinical and economic utility. Fourth, potential verification bias should be considered, as women who tested MHSA-positive but APTIMA-negative did not systematically undergo histological confirmation. Nevertheless, the likelihood of missing high-grade lesions remains very low, given that all enrolled women were managed under a co-testing strategy (combining cytology and HPV testing) and received several rounds (4 rounds for those who tested positive at baseline and 2 rounds for those who tested negative) of cytology follow-up.

This study demonstrated that both APTIMA and MHSA provide robust diagnostic performance for CIN2+ and CIN3+ detection in a clinical screening setting within 3 years. While APTIMA remains the clinically validated standard, the MHSA shows promise as an alternative HPV testing strategy, particularly in settings exploring the utility of NGS-based molecular diagnostics. With further validation and cost reduction, MHSA has the potential to evolve into a comprehensive public health platform for cervical cancer prevention and control.

SUPPLEMENTARY MATERIALS

Table S1

Comparison of HPV positivity rates between APTIMA and MHSA (n=2,683)

Table S2

CIN2 and CIN3+ detection by cytology and HPV assays at baseline and first-year follow-up

Table S3

CIN2 and CIN3+ detection by cytology and HPV assays at second and third-year follow-up

Table S4

IR and IRR for CIN2+ according to baseline test results (n=2,683)

Table S5

IR and IRR for CIN3+ according to baseline test results (n=2,683)

Fig. S1

Age-specific prevalence of 14 HR-HPV types detected by APTIMA and MHSA. Prevalence of HPV infection in different age groups using the 2 HPV detection methods. In the APTIMA group, the percentage of HPV was 21.82% in the under 30 years old group, 11.28% in the 30–40 years old group, 16.58% in the 40–50 years old group, 22.81% in the 50–60 years old group, 22.73% in the over 60 years old group. In the MHSA group, the percentage of HPV was 20.00% in the under 30 years old group, 12.31% in the 30–40 years old group, 20.84% in the 40–50 years old group, 25.83% in the 50–60 years old group, 19.70% in the over 60 years old group.

Fig. S2

Comparison of HPV positivity by histopathological diagnosis: APTIMA vs. MHSA.

Fig. S3

ROC curves for 3-year detection of CIN2+ and CIN3+ using different screening strategies. (A-D) panels illustrate the diagnostic performance of APTIMA and MHSA combined with cytology (co-test or triage) for detecting CIN2+ and CIN3+ within 3 years. (A) Performance of the APTIMA and MHSA with co - test cytology for the detection of CIN2+ within 3 years; (B) Performance of the APTIMA and MHSA with co - test cytology for the detection of CIN3+ within 3 years; (C) Performance of the APTIMA and MHSA with cytology triage for the detection of CIN2+ within 3 years; (D) Performance of the APTIMA and MHSA with cytology triage for the detection of CIN3+ within 3 years.

Fig. S4

Pre- and post-test probability plots for CIN3+ risk over 3 years. Illustrates the transformation from pretest to posttest probability of CIN3+ over a 3-year period based on likelihood ratios. (A) Fagan nomogram of MHSA combined with cytology for the CIN3+; (B) Fagan nomogram of APTIMA combined with cytology for the CIN3+; (C) Fagan nomogram of MHSA with cytology triage for the CIN3+; (D) Fagan nomogram of APTIMA with cytology triage for the CIN3+.

Fig. S5

Distribution of HPV positivity across the study population. For simplicity, HPV68a and HPV68b are listed as HR-HPV types.