N: number; CI: 95% confidence interval; CIN: cervical intraepithelial
neoplasia
+/+ positive on FVU and cervical samples, +/- positive only on cervical
samples, -/+ positive only on FVU samples, -/- negative on both sample
types.
Note: Color legend: for the concordance: dark green (1.00 ≥ K
> 0.80): excellent; light green (0.80 ≥ K >
0.60): good; yellow (0.60 ≥ K > 0.40): moderate; orange
(0.40 ≥ K > 0.20): fair; red (0.20 ≥ K >
0.00): poor.
Discussion
The introduction of self-sampling in cervical cancer screening programs,
further enhanced by the COVID19 pandemics (17), is an
important instrument to reach 70% screening coverage as proposed in the
WHO call to action (1). The similar clinical accuracy
of PCR-based HPV tests on self-samples and clinician-collected cervical
scrapes has been demonstrated in previous validation studies for other
assays (10–15).
The present study demonstrated that the use of OncoPredict HPV SCR assay
on self-collected vaginal specimens FLOQSwabs and resuspended in 5 ml of
eNat and FVU collected using Colli-Pee FV5000 has a similar clinical
accuracy to detect ≥CIN2 and CIN3 lesions as compared to
clinician-collected cervical samples. Clinical sensitivity of
OncoPredict HPV SCR assay on FVU and vaginal samples was similar to
cervical specimens, however specificity on vaginal samples was lower
when applying manufacturer cut-off values. Cut-off optimization on
vaginal self-collected samples resulted in an improvement in specificity
without compromising sensitivity. A lower specificity for the detection
of ≥CIN2 as compared to cervical specimen was also reported for the
validation of BD Onclarity HPV test on FLOQSwabs resuspended in 3 ml of
BD HPV self-collection diluent (18). On the other
hand, in the Belgian VALHUDES, where vaginal samples were resuspended in
20 ml of PreservCyt, a posteriori cut-off determination was necessary to
improve the clinical sensitivity, but not for the specificity
(11,12).
Different preanalytical workflows of self-collected vaginal samples may
affect on clinical accuracy of the test. Therefore, optimization and
standardization of the procedures for handling and testing of
self-samples are fundamental to ensure an optimal performance of the
assay (19). Presently, the VALHUDES protocol has been
developed to assess the performance of HPV tests in combination with
self-collection devices (9). In the European VALHUDES,
vaginal swabs have been collected using FLOQSwabs resuspended in 5 ml of
eNat, while FVU was collected using a 20 ml Colli-Pee device. eNat is a
transport medium that allows the preservation of nucleic acids,
denaturation of proteins and inactivation of microbial agents. It has
been previously used in combination with HPV molecular assays
(20,21);two studies demonstrated a good analytical
performance of FLOQSwabs resuspended in 5 ml of eNat as compared to
cervical samples (22,23). Moderate to excellent
agreement between vaginal and cervical specimens was also demonstrated
in the present study.
Both urine and vaginal self-collected samples are well accepted by women
(24), in particular this study confirmed that FVU is a
non-invasive collection method with clinical accuracy for ≥CIN2 lesions
comparable to cervical specimens, as previously reported
(14,15).
Ensuring sample adequacy is crucial, particularly for self-collection,
in order to prevent false-negative results (25–27).
One of the main advantages of OncoPredict HPV SCR assay is the inclusion
of a thorough quality assessment, both for the preanalytical and
analytical phases. The assay allows to determine the efficiency of
nucleic acid extraction and potential PCR inhibition through the use of
external calibrators, as well as assessing adequacy of sample collection
through a quantitative cellularity assessment. In most molecular assays
an internal housekeeping gene is used for both sample adequacy and
amplification assessment. Recent studies have underlined the importance
of identifying the cause of invalidity (25) of
performing quality controls for the assay in a separate reaction well
(26). In the present study, no invalid result related
to PCR inhibition was detected, underlying the good performance of the
analytical process. The invalidity due to a low extraction efficiency
could be attributed to errors in specific nucleic acid extraction runs
that may be resolved by retesting samples after a new extraction. On the
contrary, in case of low cellularity samples, in absence of other
invalidity reasons, sample collection should be repeated
(25). In general, the invalidity rate was higher in
cervical samples than in self-collected samples. This could be related
to different limits of acceptable cellularity for cervical and
self-collected samples. Moreover, as previously discussed, in the
present study vaginal samples were resuspended in 5 ml of eNat while
cervical swabs in 20 ml of PreservCyt which may have resulted in lower
sample cellularity. Finally, the inclusion of the QC module in
OncoPredict HPV SCR assay allows to accurately verify the sample
adequacy, avoiding false negative results in cervical cancer screening.
OncoPredict HPV SCR assay is a limited genotyping assay, identifying
HPV16, HPV18 and/or “other” hrHPV genotypes, whereas OncoPredict HPV
Quantitative Typing (QT) is a full genotyping assay that can distinguish
all the 12 hrHPV types separately. Both assays have been independently
validated on cervical and self-samples withing VALGENT and VALHUDES
frameworks (Latsuzbaia et al., submitted, 16,28),
respectively. hrHPV-positive specimens identified with OncoPredict HPV
SCR assay may benefit from complete genotyping using QT assay as a
reflex test.
In conclusion, following a posteriori cut-offs adjustment the
OncoPredict HPV SCR assay demonstrated similar clinical accuracy for
≥CIN2 lesions on self-collected vaginal and FVU samples compared to
testing on clinician-collected cervical samples.