Dear Editor,

We read with interest the recently published manuscript by Kir et al. entitled “High-risk human papillomavirus (HPV) detection in formalin-fixed paraffin-embedded cervical tissues: performances of Aptima HPV assay and Becton Dickinson (BD) Onclarity assay” [1]. The study evaluates the off-label use of commercial HPV assays on formalin-fixed paraffin-embedded (FFPE) specimens. The authors detail the performance of the Hologic Aptima RNA test and the BD Onclarity DNA test in a study of 189 cases (46 SCC, 107 HSIL and 36 benign/normal). They report that, while the specificity and positive predictive value (PPV) were 100% for both assays, the Aptima assay was more sensitive, detecting 99.4% (95% CI 96.46% to 99.98%) of CIN2+ cases, versus the BD Onclarity assay with a sensitivity of 75.9% (95% CI 65.27% to 84.62%) [1]. The authors conclude that “both assays are reliable methods for high-risk HPV detection and genotype determination in FFPE specimens” and that the “Aptima assay has the advantage of higher sensitivity”.

We believe that there are a number of deficiencies in the study design (not addressed in the discussion section) which call into question the validity of the conclusions. Firstly, this is not a split sample study, nor a true head to head comparison since the samples were not randomly assigned to each assay for testing. All 189 cases were first tested with the Aptima assay and only approximately half (n = 97) were subsequently tested with the BD Onclarity assay. Kir et al state that after collection of the Aptima samples H&E tissue staining was used to check to see if there was sufficient lesion remaining for BD Onclarity testing and they found that the lesion was depleted in 92, or almost half of the total available cases. The disease profile of the 97 remaining cases was SCC: n=40, 41.2%; HSIL: n=43, 44.3% and benign/normal: n=14, 14.4% which is different from that of the full cohort SCC: n=46, 24.3%; HSIL: n=107, 56.6%; benign/normal: n=36, 19.0%. Thus, the Aptima assay had the opportunity to evaluate 153 presumptive HPV positive cases, versus just 83 cases for Onclarity. The loss of almost half of the total samples to lesion depletion is surprising and can be explained by how the samples were processed. Firstly, 4 μm thick sections were cut from the block for separate immuno-staining with p16 and Ki67 to confirm both SCC and HSIL diagnoses. Thereafter, “from each tissue block, 8–60 (depending on tissue size) 4 μm thick sections” were harvested for Aptima HPV testing. The use of a minimum of 8, and as many as 60, sections for nucleic acid extraction is excessive in our experience. It is therefore not unexpected that the use of multiple sectioning extended past the lesion, rendering the sample unsuitable for further testing. The authors do not detail exactly how they used H&E to confirm lesion size/sample adequacy prior to BD Onclarity testing. Best practice is to use a sandwich method where both sides of the section(s) used for nucleic acid extraction are confirmed to have visible lesions by H&E staining [2]. The loss of ~50% of samples due to upstream sampling, where a minimum of 10 preceding sections were harvested, suggests that the residual samples tested by BD Onclarity were not of the same quality in terms of lesion size, and hence viral load. Moreover, Kir et al appear to have used just a single 5-10 micrometer section for BD Onclarity testing versus 8-60 sections for Hologic Aptima test. While this may reflect the different sample input requirements for each assay, it does underline the inequality in the workflows in terms of target concentration (and calls into question the assertion that the BD Onclarity assay is less sensitive). A simpler and improved study design would have been to randomly assign one of the 8-60 Aptima sections for testing with BD Onclarity assay. Our experience with FFPE tissue is that there is ample cellularity in a single section and that HPV is detected at high frequency provided there is a lesion present. This is evident from analysis of average beta-globin cellularity control Ct scores from both liquid-based-cytology (LBC) and single FFPE sections, which have a similar (normal) and overlapping distribution [3].
Kir et al. cite a prior BD Onclarity FFPE study where the sensitivity for CIN2+ was 90% [4] which compares favorably with the Aptima performance in their study and those of other assays in previously published work (Table 5) [1]. Two additional BD Onclarity studies (not cited in Table 5) investigated the performance of the assay using FFPE sections: Genta et al reported 84% HPV positivity in 292 cases of invasive cervical cancer [5] and Bottari et al. recently compared the performance in 99 FFPE samples to paired LBC specimens (26 CIN 1, 30 CIN 2, and 43 CIN 3+). After 15 samples were excluded due to sample quality, the remaining 84 samples recorded an overall agreement of 89% for HPV status between FFPE Onclarity samples versus LBC samples [6]. (It is also noteworthy that all three published Onclarity studies used the gold standard sandwich method to ensure that the lesion was present).

The afore-mentioned studies suggest that the performance of the BD Onclarity assay using FFPE specimens is similar to other published assays, and we respectfully submit that the apparent reduction in sensitivity in assay performance reported by Kir et al is likely due to sample quality/viral load differences between the Aptima and BD Onclarity samples. Finally, the authors interpret the higher sensitivity of the Aptima assay to mean that “choosing RNA [vs. DNA] for nucleic acid extraction from FFPE tissue will lead to better results”. This seems at odds with the known lability of RNA versus DNA [7] and is not in fact supported by the published work they cite to support this assertion [8]. The increased “sensitivity” is more likely to have resulted from the increased sample input volume (as much as 60-fold higher) for the RNA versus DNA assays compared in the study.

Sincerely,

REFERENCES

1. Kir G, Gunel H, Olgun ZC, McCluggage WG: High-risk human papillomavirus (HPV) detection in formalin-fixed paraffin-embedded cervical tissues: performances of Aptima HPV assay and Beckton Dickinson (BD) Onclarity assay. Journal of clinical pathology 2021.
2. Mena M, Lloveras B, Tous S, Bogers J, Maffini F, Gangane N, Kumar RV, Somanathan T, Lucas E, Anantharaman D et al: Development and validation of a protocol for optimizing the use of paraffin blocks in molecular epidemiological studies: The example from the HPV-AHEAD study. PloS one 2017, 12(10):e0184520.
3. Maus C, Vaughan, L, Adams M, Chen, C, Dixon, E, Gutierrez, E, Harris, J, Horlick, E, Leitch, S, McMillian, R, Murphy, P, Nelson, R., Nussbaumer, W, Peck, J, Porter, M, Richart, G, and Mertz, L.: Performance Of BD Viper™ HPV Assay Using Formalin-Fixed Paraffin-Embedded (FFPE) Samples. EUROGIN Conference Oral Presentation, Lisbon, Portual May 8-11 2011.
4. Castro FA, Koshiol J, Quint W, Wheeler CM, Gillison ML, Vaughan LM, Kleter B, van Doorn LJ, Chaturvedi AK, Hildesheim A et al: Detection of HPV DNA in paraffin-embedded cervical samples: a comparison of four genotyping methods. BMC infectious diseases 2015, 15:544.
5. Nogueira Dias Genta ML, Martins TR, Mendoza Lopez RV, Sadalla JC, de Carvalho JPM, Baracat EC, Levi JE, Carvalho JP: Multiple HPV genotype infection impact on invasive cervical cancer presentation and survival. PloS one 2017, 12(8):e0182854.
6. Bottari F, Passerini R, Renne G, Guerrieri ME, Sandri MT, Li A, Orlandini A, Iacobone AD: Onclarity Performance in Human Papillomavirus DNA Detection in Formalin-Fixed Paraffin-Embedded Cervical Samples. Journal of lower genital tract disease 2021, 25(3):216-220.
7. Groelz D, Viertler C, Pabst D, Dettmann N, Zatloukal K: Impact of storage conditions on the quality of nucleic acids in paraffin embedded tissues. PloS one 2018, 13(9):e0203608.
8. Ferrer I, Armstrong J, Capellari S, Parchi P, Arzberger T, Bell J, Budka H, Ströbel T, Giaccone G, Rossi G et al: Effects of formalin fixation, paraffin embedding, and time of storage on DNA preservation in brain tissue: a BrainNet Europe study. Brain pathology (Zurich, Switzerland) 2007, 17(3):297-303.