Point-of-care hrHPV mRNA detection with reverse transcription recombinase polymerase amplification on a portable fluorimeter from provider-collected cervical samples in Maputo, Mozambique Authors: Emilie Newsham Novak, Cannon J. Hansen, Zoha Wazir, Ariel Ma, Jennifer Carns, Hira Atif, Jane R. Montealegre, Michael E. Scheurer, Cesaltina Lorenzoni, Daniel G. Rosen, Mila P. Salcedo, Kathleen M. Schmeler, Rebecca R. Richards-Kortum Published: 2026 ABSTRACT BackgroundCervical cancer is the most common cancer in women in Mozambique, where limited screening infrastructure contributes to high disease burden. While testing for high-risk human papillomavirus (hrHPV) DNA can identify women at risk for cervical cancer, its limited specificity may lead to overtreatment in screen-and-treat programs, straining resources. We performed a pilot study to evaluate preliminary feasibility of a novel, accessible test to detect hrHPV mRNA, a more specific biomarker for cervical cancer risk, from 22 women who previously screened positive for visual inspection with acetic acid at Maputo Central Hospital in Mozambique.MethodsCervicovaginal samples were analyzed for hrHPV DNA using the Xpert HPV test. RNA was extracted from samples using a benchtop spinner instead of a centrifuge. Isothermal reverse transcription recombinase polymerase amplification (RT-RPA) assays targeting HPV 16, 18, and 45 mRNA, plus a cellular control were performed on a portable fluorimeter with results compared to the gold standard portable reverse transcription quantitative polymerase chain reaction (RT-qPCR). Cervical biopsies were obtained and submitted for histopathologic analysis.ResultsRT-RPA and RT-qPCR results were in 100% agreement (22/22). The sample-to-answer times were 1.5 h for the RT-RPA test and 3.75 h for RT-qPCR. All RT-RPA-positive samples were Xpert-positive for the same HPV type. Two Xpert-positive samples were RT-RPA-negative; four were positive for high-risk HPV types other than 16, 18, or 45. RT-RPA had a higher positive predictive value (100%) for high grade lesions by biopsy than Xpert (73%). Xpert had a higher negative predictive value (82%) than RT-RPA (71%) for high grade lesions.ConclusionsThis study demonstrates the potential feasibility of novel point-of-care hrHPV mRNA testing in a low-income country. Additional work is needed before this test is clinically relevant and deployable in resource-limited settings.Trial registrationThis study was registered on 2022-04-25 on clinicaltrials.gov under registration number NCT05372484.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13027-026-00772-5. Background In Mozambique, cervical cancer is the most common cancer in women [1]. The high burden from cervical cancer in Mozambique is largely attributed to low human papillomavirus (HPV) vaccination rates and a lack of sufficient screening and treatment services [2]. Mozambique also has high rates of human immunodeficiency virus (HIV) infection, a significant risk factor for cervical cancer [3, 4]. The World Health Organization (WHO) recommends testing for high-risk human papillomavirus hrHPV DNA to screen for cervical cancer and its precursors [5]. This kind of testing, however, is expensive and requires advanced healthcare and laboratory infrastructure, making it challenging to implement in low-income countries [6, 7]. In their 2024 report, Unitaid found 14 hrHPV nucleic acid amplification tests that were both commercially available and clinically validated; however, only a small subset of these tests are feasible in low and middle-income countries (LMICs) [8, 9]. While hrHPV DNA is highly sensitive for cervical cancer, it has low specificity for high-grade lesions – defined as cervical intraepithelial neoplasia 2 or higher (CIN 2+) – as most hrHPV infections are transient and clear naturally without intervention [10, 11]. In screen-and-treat programs where point-of-care hrHPV testing is most needed, this results in high rates of overtreatment, straining limited resources and causing potential unnecessary anxiety, discomfort and complications [12]. A more specific biomarker is hrHPV E6/E7 mRNA, which is overexpressed in hrHPV infections where the viral DNA has integrated into the patient’s DNA, making cervical precancer more likely [13]. For this reason, hrHPV mRNA is a more specific biomarker for high-grade cervical precancer, with a reported positive predictive value (PPV) of 78% and a high negative predictive value (NPV) of (99%) [10]. Although hrHPV DNA testing also has a high NPV (99%), its PPV is only 15–25% [10, 14]. While neither biomarker is a perfect predictor of high-grade precancer, the higher PPV of mRNA could make it preferable in circumstances where high specificity is needed, as supported by multiple systematic reviews and meta-analyses [15–17]. Current methods of hrHPV mRNA testing are even more challenging to implement than DNA testing methods due to the need for more complicated and time-intensive extraction and amplification methods. In a recent Unitaid report, only one mRNA test (Hologic’s Aptima mRNA test) was reported to be commercially available and clinically validated, and in LMICs it is only feasible in centralized labs and not at the point of care [6, 8]. Resource-limited settings (RLS) with a high cervical cancer burden would benefit from an accessible hrHPV mRNA test that could provide more specific cervical cancer screening results in a timely manner at the point of care [18]. In this manuscript, seek to evaluate whether a point-of-care mRNA test that employs isothermal amplification and detection on a portable fluorimeter can be successfully performed in a RLS like Mozambique. We demonstrate detection of hrHPV mRNA at the point of care from cervicovaginal samples collected from 22 patients at risk for cervical cancer at Maputo Central Hospital in Mozambique, with the hypothesis that hrHPV mRNA would have a higher positive predictive value for high-grade cervical precancer than hrHPV DNA. We use a simplified protocol to extract RNA from freshly collected cervicovaginal samples and a real-time, isothermal amplification method on a small, portable fluorimeter to rapidly detect mRNA from the three most carcinogenic HPV types (16, 18, and 45). We compared results of the isothermal amplification method to the gold standard of reverse transcription quantitative polymerase chain reaction (RT-qPCR) using a portable RT-qPCR machine. We also tested the 22 samples for hrHPV DNA using the Xpert HPV test. Finally, we compared the hrHPV results to pathology results from biopsies collected from each patient. Methods Reagent transportation from Houston to Mozambique Thermosensitive reagents for reverse transcription quantitative polymerase chain reaction (RT-qPCR) and reverse transcription recombinase polymerase amplification (RT-RPA) were transported from Houston to Mozambique in a thermos (4AllFamily) with an ice pack via carry-on luggage. Reagent stability was assessed before travel by placing reagents in the thermos with the ice pack for 48 h and running the RT-qPCR and RT-RPA assays described previously [19] at 0-, 24-, and 48-hour timepoints with fresh extracted RNA. Reagents were stored in a -20 °C freezer upon arrival in Mozambique. Patient enrollment and sample collection Patients attending cervical cancer screening clinics at community hospitals around Mozambique who had an intact cervix, were not pregnant, and who screened positive with visual inspection with acetic acid were eligible to participate in this study. The study nurses reached out to nurses in the cervical cancer screening and HIV clinics within the primary healthcare facilities in the community to identify potential eligible subjects. Participants provided written informed consent. The study was approved by the institutional review boards (IRBs) at Rice University, MD Anderson Cancer Center, Baylor College of Medicine, and the Committee for Bioethics in Health of Mozambique, and is registered on clinicaltrials.gov (NCT05372484). 22 consecutively enrolled subjects from June 3, 2025 to June 7, 2025 were included in the study. A provider-collected cervicovaginal sample was taken using a cytobrush (Kashi Scientific) and swirled in 20 mL of PreservCyt (ThinPrep) media for 30 s. Colposcopy was performed and all areas considered abnormal were subject to biopsy, at the discretion of the healthcare provider. If no abnormalities were noted, a random biopsy was performed from a normal site at the squamous columnar junction (SCJ). Modified RNA extraction The protocol for the Monarch Spin RNA Isolation Kit (New England Biolabs) was modified for use with a quick spinner (VWR) instead of a high-speed programmable benchtop microcentrifuge, as recommended in the protocol. First, a 1 mL aliquot was removed from each well-mixed sample in PreservCyt. Cells from this aliquot were pelleted by sedimentation overnight or by spinning in a quick spinner for 3–4 min. After pelleting, the instructions provided with the kit were followed with the modification that all spin times were doubled from 1 to 2 min, except for the elution step, which was tripled from 30 s to 3 min. A more detailed description of this process can be found in the Supplementary Information. Following elution into 50 µL water, 1.5 µL DNase I, 15 µL DNase I reaction buffer (New England Biolabs), and 83.5 µL water were added and incubated at 37 °C for 30 min. After incubation, the enzymes were deactivated at 75 °C for 10 min and the sample was placed in the refrigerator for immediate use or in a -20 °C freezer for longer-term (> 2 h) storage. The total extraction time was approximately 75 min, including hands-on and instrumentation time. RT-RPA with portable fluorimeter Reverse transcription recombinase polymerase amplification (RT-RPA) reactions for HPV 16, HPV 18, HPV 45, and a β-actin cellular control were performed with extracted RNA using reaction conditions described previously (n = 1 for each reaction per sample) [19]. Reactions were run on the Axxin T8-ISO fluorimeter and analyzed using a MATLAB code with baseline subtraction (See Supplementary Information). Results were obtained in 15 min or fewer of instrumentation time. RT-qPCR with portable thermocycler Reverse transcription quantitative polymerase chain reaction (RT-qPCR) reactions for HPV 16, HPV 18, HPV 45, and a β-actin cellular control were performed with extracted RNA using reagents and reaction conditions described previously (n = 1 for each reaction per sample) [19]. This test was chosen as a reference test because it was the best gold standard that could be transported to Maputo. No-target controls, HPV 16, HPV 18, and HPV 45 gBlocks, genomic DNA, and total RNA extracted from SiHa, HeLa, and MS751 cells (ATCC) were tested before patient samples were run to confirm test validity. Reactions were run on the PreciGenome FAST-16-3 mini portable PCR instrument with a threshold of 3000 RFU. Any cycle threshold (Ct) higher than 39 was considered negative based on the limit of detection of this primer set. Results were obtained in approximately 2.5 h of instrumentation time. RT-qPCR results were obtained after RT-RPA results and were therefore not available to the operator of the RT-RPA test, but the RT-RPA results were known prior to the RT-qPCR results. Xpert HPV DNA testing Xpert HPV (Cepheid) tests were run for each sample to determine the presence of HPV 16, HPV 18/45, or other high-risk HPV DNA. Pathology Biopsy samples were fixed in formalin, embedded in paraffin blocks, and sectioned using a microtome. Tissue sections were stained using hematoxylin and eosin. Biopsy slides were reviewed and assessed using standard histopathologic criteria by pathologists in Maputo, Mozambique according to the WHO blue book [20]. Each slide was reviewed by a minimum of two pathologists who reported a consensus diagnosis, and pathologists were blinded to the molecular results. Data analysis The predictive values of the tests for high-grade cervical lesions were reported in terms of NPV and PPV. The Wilson-Brown method was used to calculate 95% confidence intervals. Performance metrics such as sensitivity and specificity were not calculated due to the small sample size. Results Summary of patient data HIV status, Xpert HPV, RT-qPCR, RT-RPA and pathology results for all 22 patients are summarized in Table S1. Sixteen participants were living with HIV. Seven patients tested positive for HPV 16/18/45 DNA and seven tested positive for other hrHPV DNA types by Xpert HPV. Ten patients had biopsies with cervical intraepithelial neoplasia 2 (CIN 2) or higher-grade pathology, including one patient with squamous cell carcinoma and one patient with adenocarcinoma. Performance of RT-RPA mRNA test compared to RT-qPCR gold standard As shown in Fig. 1, of the 22 samples tested, four were HPV 16 mRNA-positive, one was HPV 18 mRNA-positive, and 17 were negative for HPV 16/18/45 mRNA, with valid amplification of the cellular control. Results of the RT-RPA test were in 100% agreement with the RT-qPCR mRNA test (Table S2). Of note, while an HPV 45 test was run for all the samples, none of the samples were HPV 45-positive by either test. Fig. 1Time to result for RT-RPA and RT-qPCR tests. (A) Time to positivity (minutes) for HPV 18, HPV 16, and ACTB RT-RPA tests run for each sample. (B) Cycle threshold (Ct) for HPV 18, HPV 16, and ACTB RT-qPCR tests run for each sample Time to result for RT-RPA and RT-qPCR tests. (A) Time to positivity (minutes) for HPV 18, HPV 16, and ACTB RT-RPA tests run for each sample. (B) Cycle threshold (Ct) for HPV 18, HPV 16, and ACTB RT-qPCR tests run for each sample Predictive value of hrHPV RNA and DNA tests for high-grade cervical lesions The RT-RPA RNA test accurately identified five high-grade cervical lesions (CIN 2+), while five RT-RPA-negative results were CIN 2+ (Table 1). All biopsies with pathology < CIN 2 were associated with RT-RPA-negative results. The RT-RPA assay had a PPV of 100% and NPV of 71% (Table 2). The Xpert HPV DNA test accurately identified eight high-grade cervical lesions, with three Xpert-positive samples corresponding to pathology results < CIN 2 (Table 1). Two Xpert-negative results had pathology results of CIN 2+, and nine Xpert-negative results had pathology results showing < CIN 2 (Table 1). The Xpert HPV test had a PPV of 73% and an NPV of 82% (Table 2). Results of the Xpert HPV DNA test show that two patients with CIN 2 + pathology were infected with hrHPV types other than HPV 16/18/45. Since these additional types are not part of the RT-RPA test being evaluated, we also assessed the performance of RT-RPA after excluding these samples. This is not a clinically relevant exclusion but is helpful in evaluating test performance. Excluding these results, the RT-RPA test has a PPV of 100% and an NPV of 77% (Table 2). Table 1Results of mRNA (left) and DNA (right) tests compared to pathologyRT-RPA mRNACIN 2+< CIN 2TotalXpert HPV DNACIN 2+< CIN 2TotalmRNA+505DNA +8311mRNA-51217DNA-2911Total101222Total101222 Table 1Results of mRNA (left) and DNA (right) tests compared to pathologyRT-RPA mRNACIN 2+< CIN 2TotalXpert HPV DNACIN 2+< CIN 2TotalmRNA+505DNA +8311mRNA-51217DNA-2911Total101222Total101222 Results of mRNA (left) and DNA (right) tests compared to pathology Table 2PPV and NPV with 95% confidence interval (CI) of mRNA and DNA tests for high-grade pathology (CIN 2+)TestPositive predictive value (95% CI)Negative predictive value (95% CI)mRNA RT-RPA100% (57–100%)71% (47–87%)Xpert HPV73% (43–90%)82% (52–97%)mRNA RT-RPA (excluding samples positive for hrHPV DNA types not included in the RT-RPA assay)100% (57–100%)77% (50–92%) Table 2PPV and NPV with 95% confidence interval (CI) of mRNA and DNA tests for high-grade pathology (CIN 2+)TestPositive predictive value (95% CI)Negative predictive value (95% CI)mRNA RT-RPA100% (57–100%)71% (47–87%)Xpert HPV73% (43–90%)82% (52–97%)mRNA RT-RPA (excluding samples positive for hrHPV DNA types not included in the RT-RPA assay)100% (57–100%)77% (50–92%) PPV and NPV with 95% confidence interval (CI) of mRNA and DNA tests for high-grade pathology (CIN 2+) Discussion The methods and results described in this study constitute a proof-of-concept pilot study with several strengths that may eventually contribute significantly to preventing cervical cancer in RLS like Mozambique, a country with a high burden of cervical cancer. Most notably, the entire sample-to-answer process for hrHPV mRNA testing by both the more accessible RT-RPA test and the gold standard RT-qPCR test were performed onsite at the hospital in which patients were being seen, with same-day results. In clinical practice, this point-of-care testing would permit timely decisions regarding patient care, especially in the absence of histopathology and when loss to follow-up is a major concern. Second, because all patients received a biopsy, comparing the RT-RPA results to both RT-qPCR and pathology permits assessment of both accurate hrHPV mRNA detection and whether mRNA detection corresponds to clinically significant disease. Finally, the inclusion of Xpert HPV permits extended genotyping of all hrHPV infections and demonstrates the higher PPV of hrHPV mRNA than hrHPV DNA in this study. There are several limitations to this study. Most significantly, the small sample size limits the broad generalization of the hrHPV mRNA, DNA, and pathology results to broad patient populations, and the three HPV mRNA types evaluated are responsible for only approximately 75% of cervical cancers [21]. Additionally, the reagents and techniques employed will require further optimization before this test is accessible for use at the point of care. The modified extraction technique, for example, takes approximately 75 min, and the test requires many user steps to run, which is likely infeasible for laboratory personnel in a resource-limited clinic. Importantly, in this study, the test was run by laboratory personnel with a high level of training, and all the equipment used in the study was brought to the lab from Houston, Texas. Usability metrics, operator-to-operator variability, day-to-day performance, and robustness under different environmental conditions were not assessed. At present, the lab at Maputo Central Hospital does not have the infrastructure to support this kind of test in its current form. Reagents were mostly in liquid form and were transported to Mozambique, stored in a refrigerator or freezer, and used in a short amount of time with many redundant aliquots, which is inefficient, costly, and infeasible for long-term implementation. Of note, while MATLAB was used to analyze the RT-RPA results, it is not a requirement for running the test, as the Axxin T8 both displays fluorescence in real time and can be programmed to report positive vs. negative results on the machine itself based on user-defined thresholds. The concordance between the RT-RPA and RT-qPCR tests indicates that the RT-RPA test can accurately detect hrHPV mRNA at the point of care. This could make hrHPV testing more accessible in RLS, as RT-RPA is significantly faster (15–20 min compared to ~ 2 h) and requires less instrumentation than RT-qPCR. Furthermore, the 100% validity rate of both tests highlights their ability to be successfully run at the point of care in a setting with limited laboratory infrastructure using portable instruments and point-of-care-friendly techniques. Of note, because there were no HPV 45-positive samples, we can only assess the successful detection of HPV 16 and HPV 18 mRNA. Additionally, while β-actin was run as a sample adequacy control, no spectrophotometer, Bioanalyzer, Qubit, or gel electrophoresis equipment was available for gold standard RNA integrity measurements (such as A260/280 or RNA integrity number). Finally, technical replicates could not be performed due to the low throughput of the machines and limited time available in the hospital for testing, preventing evaluations of the precision and reliability of the results. When assessing the association between hrHPV tests and CIN 2 or higher pathology, the Xpert HPV DNA test had a higher NPV (82%) than the RT-RPA mRNA test (71%), with only two false negatives. The RT-RPA test, however, had a higher PPV for high-grade cervical precancer and cancer (100%), with zero false positives, suggesting, with further validation, its potential as a triage tool when specificity to high-grade pathology is of elevated importance. This is a key benefit of this test, and we hope that in the future, it could be used to reduce overtreatment in screen-and-treat programs. The low NPV of the RT-RPA test can be partly attributed to its limited hrHPV genotypes [3] compared to the Xpert HPV test’s 14 hrHPV genotypes. When false negative RT-RPA results were excluded because the Xpert HPV result is an hrHPV type that the RT-RPA test would not detect, the NPV of the RT-RPA test increased to 77%. We acknowledge that this does not make the test sufficient for use in clinical practice; however, it is promising to know that more accurate results might be achievable with extended mRNA genotyping capabilities. Notably, hrHPV mRNA is not a perfect biomarker for high-grade precancer – even in commercial RNA tests that cover more hrHPV genotypes, the NPV and PPV for CIN 2 + pathology are reported to be 90.8% and 52.8%, respectively for PreTect HPV Proofer and 96.4% and 32.3% for Aptima HPV [22, 23]. When testing for hrHPV mRNA, the WHO recommends five-year screening intervals, as opposed to ten-year intervals for hrHPV DNA due to uncertain longitudinal evidence on hrHPV mRNA test performance. Furthermore, because a referral population of only VIA-positive women was tested, the PPV calculated here is likely higher than it would be for a screening population. Finally, the clinical impact of these tests cannot be assessed as they were not used to inform patient care. This is a rare study of hrHPV mRNA testing done in a population with a high rate of women living with HIV (WLWH) (73%). A few studies have demonstrated hrHPV mRNA testing using commercially available platforms in low and middle-income countries in a high-prevalence HIV population [18, 24–27], but additional studies are needed to validate the efficacy of mRNA as a biomarker for high-grade precancer in WLHIV. It is possible that this aspect of the study may contribute to the low sensitivity reported, since HIV coinfection makes hrHPV infections more likely to cause high-grade cervical precancer [28, 29]. Notably, the WHO does not recommend hrHPV mRNA testing in WLHIV due to limited data assessing its efficacy as a biomarker in this population [30]. Larger studies are necessary to determine whether mRNA can retain sufficient sensitivity to be useful for WLHIV. Nevertheless, this study demonstrated successful detection of hrHPV mRNA from samples with HIV co-infections. Additional work is needed before the RT-RPA test for hrHPV mRNA described here is clinically relevant and deployable in RLS. Most importantly, five additional genotypes (33, 58, 31, 52, and 35) are necessary to capture a more complete range of hrHPV infections that are likely to cause cervical cancer [9]. The test would also be much more accessible if an extraction-free sample preparation protocol were incorporated to shorten the time to result and reduce the burden of labor on laboratory personnel. Specifically, a heat or chemical-based extraction-free mRNA sample preparation process would greatly reduce the number of user steps prior to RT-RPA amplification [19, 31]. More thermostable reagents, multiplexing of multiple hrHPV genotypes into a single tube, and other infrastructure and cost-reducing measures would also improve the accessibility and feasibility of running this test in settings like Mozambique. Conclusions Overall, this study demonstrates a proof-of-concept evaluation of point-of-care hrHPV mRNA testing in a low-income country like Mozambique. The data presented here will be valuable for informing future, larger studies of the feasibility and clinical relevance of hrHPV mRNA testing. Future iterations of the test will focus on making it more point-of-care-friendly and clinically useful with larger sample sizes, additional hrHPV genotypes, and simplified mRNA sample preparation. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1