Self-collection of vaginal specimens for human papillomavirus testing in cervical cancer prevention (MARCH): a community-based randomised controlled trial.

Background: Vaginal self-sampling for human papillomavirus (HPV) DNA testing could increase rates of screening participation. In clinic-based settings, vaginal HPV testing is at least as sensitive as cytology for detecting cervical intraepithelial neoplasia (CIN) grade 2 or worse; however, effectiveness in home settings is unknown. We aimed to establish the relative sensitivity and positive predictive value for HPV screening of vaginal samples self-collected at home as compared with clinic-based cervical cytology.

Methods: We did a community-based, randomised equivalence trial in Mexican women of low socioeconomic status aged 25-65 years. Participants came from 540 medically underserved, predominantly rural communities in Morelos, Guerrero, and the state of Mexico. Our primary endpoint was CIN 2 or worse, detected by colposcopy. We used a computer-generated randomisation sequence to randomly allocate patients to HPV screening or cervical cytology. Eight community nurses who were masked to patient allocation received daily lists of the women's names and addresses, and did the assigned home visits. We referred women with positive results in either test to colposcopy. We did per-protocol and intention-to-screen analyses. This trial was registered with the Instituto Nacional de Salud Pública, Mexico, INSP number 590.

Findings: 12,330 women were randomly allocated to HPV screening and 12,731 to cervical cytology; 9202 women in the HPV screening group adhered to the protocol, as did 11,054 in the cervical cytology group. HPV prevalence was 9·8% (95% CI 9·1-10·4) and abnormal cytology rate was 0·38% (0·23-0·45). HPV testing identified 117·4 women with CIN 2 or worse per 10,000 (95·2-139·5) compared with 34·4 women with CIN 2 or worse per 10,000 (23·4-45·3) identified by cytology; the relative sensitivity of HPV testing was 3·4 times greater (2·4-4·9). Similarly, HPV testing detected 4·2 times (1·9-9·2) more invasive cancers than did cytology (30·4 per 10,000 [19·1-41·7] vs 7·2 per 10,000 [2·2-12·3]). The positive predictive value of HPV testing for CIN 2 or worse was 12·2% (9·9-14·5) compared with 90·5% (61·7-100) for cytology.

Interpretation: Despite the much lower positive predictive value for HPV testing of self-collected vaginal specimens compared with cytology, such testing might be preferred for detecting CIN 2 or worse in low-resource settings where restricted infrastructure reduces the effectiveness of cytology screening programmes. Because women at these sites will be screened only a few times in their lives, the high sensitivity of a HPV screen is of paramount importance.

Funding: Instituto Nacional de Salud Pública, the Health Ministry of Mexico, QiAGEN Corp.