Correction: Clinical validation of full HR-HPV genotyping HPV Selfy assay according to the international guidelines for HPV test requirements for cervical cancer screening on clinician-collected and self-collected samples

Correction: Journal of Translational Medicine (2022) 20:231 https://​doi.​org/​10.​1186/​s12967-022-03383-x

Following publication of the original article [1], we have been notified that the below-mentioned Correction article text should be published.

“We recently carried out a clinical validation of full high-risk HPV genotyping HPV Selfy assay [1] according to the international guidelines for DNA HPV test requirements for cervical cancer screening [2] as well as according to the reference protocol for validation of HPV assays for cervical cancer screening on self‑collected samples, i.e. the VALHUDES protocol [3], and we concluded that: (i) HPV Selfy is a high-risk HPV DNA test clinically validated for cervical cancer screening according to Meijer’s guidelines; and (ii) the performance of HPV Selfy on self-collected vaginal samples was non-inferior to the performance obtained on clinician-collected cervical specimens, implying that HPV Selfy is a test clinically validated for cervical cancer screening on self-collected samples.

Our paper demonstrated the non-inferiority of the performance of HPV Selfy obtained on self-collected vaginal samples applying the non-inferiority test recommended by the VALHUDES protocol [4] and, unfortunately, a statistical error has been made.

We noticed that we inverted two numbers applying the recommended formula to assess the non-inferiority of the sensitivity and specificity recorded on self-collected vaginal samples leading to incorrect results, as noticed by Marc Arbyn et al. in their letter to the Editor [5]. In particular, the number of subjects with discordant “self+ /clinician−” and “self−/clinician+” results (b and c cells in Table 4 of the paper) in the recommended formula for comparison of matched proportions were switched. Marc Arbyn et al. have claimed that the error, if corrected, would yield p values > 0.05, implying the inferiority of the specificity of HPV Selfy assay obtained on self-collected vaginal samples. Marc Arbyn et al. in their letter to the Editor stated that correct data entry would have generated non-inferiority p values 0.35 and 0.81 for sensitivity and specificity, respectively, and included in their letter to the Editor the detailed computation of the relative specificity. However, the recalculation of the relative specificity made by Marc Arbyn et al. was flawed by a statistical error: they used a δ₀ equal to 0.98 instead of 0.95 yielding reported T equal to − 0.86 that implies p value > 0.05. The correct calculation using δ₀ equal to 0.95 would quantify T equal to 2.11 implying p value < 0.05. This leads to the conclusion that still, relative specificity on self-samples is indeed significantly non-inferior to that observed in paired clinician-collected samples.

By reviewing the aforementioned calculations, we also noticed that a wrong test cut-off for sample positivity assessment has been applied while interpretating the results obtained on the self-collected vaginal samples. By applying the correct positivity cut-off for each sample type, we obtained the correct Table 4, the correct Results section of the Abstract and the correct paragraph titled “Clinical validation of HPV Selfy on self‑collected vaginal samples (VALHUDES)” that appear below with text amendments highlighted in bold italic. Additionally, there was a typo in Table 7 (a column was duplicated twice), this has been corrected; the correct Table 7 appears below.

In conclusion, we regret any inadvertent errors in our paper and we state that the corrections made allow to confirm the scientific conclusions of the article and, so, that the performance of HPV Selfy on self-collected vaginal samples was non-inferior to the performance obtained on clinician-collected cervical specimens and, consequently, that HPV Selfy is a test clinically validated for cervical cancer screening on self-collected samples. The original article has been updated.

HPV Selfy clinical sensitivity and specificity resulted non-inferior to those of HC2. By analysis of a total of 889 cervical liquid-based cytology samples from a screening population, of which 98 were from women with CIN2+, HPV Selfy showed relative sensitivity and specificity for CIN2+ of 0.98 and 1.00 respectively (non-inferiority score test: P = 0.01747 and P = 0.00414, respectively); the test reached adequate intra- and inter-laboratory reproducibility.

Next, we aimed at evaluating HPV Selfy performance on self-collected samples, as indicated by VALHUDES protocol. Hence, we needed to assess whether HPV testing on vaginal self-samples was as accurate as HPV testing on a cervical sample taken by a clinician. To do so, we identified 119 CIN2+ cases (age 25–65 years) and 791 ≤ CIN1 cases, for which we had available paired cervical specimen and self-collected vaginal samples.

HPV Selfy testing in self-collected samples was found similarly sensitive (93/96; relative sensitivity 0.97; 95% CI 0.90–1.04) and specific (722/745; relative specificity 0.97; 95% CI 0.95–0.99) to detect CIN2+ in the total study population (Table 4), in comparison with HPV Selfy performed on paired ThinPrep. Thus, HPV Selfy assay fulfills VALHUDES requirements for use of HR-HPV DNA tests on self-collected samples according to non-inferiority analysis (relative sensitivity > 0.90 with T = 1.70, p = 0.044; relative specificity > 0.95 with T = 1.87, p =0.031).

Secondary objectives of VALHUDES protocol include the assessment of the absolute accuracy of HR-HPV DNA test applied according to the sampling device and the proportion of adequate samples as determined by amplification of an internal control (a ubiquitous human gene). HPV Selfy assay provides a human beta-globin internal control, used to evaluate sample quality. In the whole study cohort, mean Ct value for the human beta-globin internal control for the HPV Selfy test on self-collected samples, obtained with the direct analysis on self-collected samples, was 26.1 Ct (median value 25.9 Ct, maximum 30.7 Ct, minimum 16.5 Ct). In the subgroup of women with biopsy-diagnosis of cervical lesions CIN2+, the same analysis resulted in 26.7 Ct (median value 26.5 Ct, maximum 30.7 Ct, minimum 24.2 Ct). This means that all women were able to self-collect a similar amount of specimen, confirming self-collected samples’ quality adequacy and so the easiness of the self-sampling procedure using the FLOQSwabs® (Copan, Brescia, Italy).