Stability of human papillomavirus (HPV) in cervical ThinPrep specimens previously lysed with glacial acetic acid: Effect on cobas 4800 HPV test performance

Authors

  • Mary McMenamin PhD,

    Corresponding author
    1. Cytopathology Department, Altnagelvin Hospital, Western Health and Social Care Trust, Londonderry, Northern Ireland, United Kingdom
    • Corresponding author: Mary McMenamin, PhD, Cytopathology Department, Altnagelvin Hospital, BT476SB, Western Health and Social Care Trust, Londonderry, Northern Ireland; Fax: (011) 02871611349; mary.mcmenamin2@westhealth.n-i.nhs.uk

    Search for more papers by this author
  • Michael McKenna MD

    1. Cytopathology Department, Altnagelvin Hospital, Western Health and Social Care Trust, Londonderry, Northern Ireland, United Kingdom
    Search for more papers by this author

  • We thank the staff from the Cytology Department, Altnagelvin Hospital for their support with this project and Mike Stevenson at Queen's University Belfast for statistical analysis of the data.

Abstract

BACKGROUND

The authors previously demonstrated that lysing samples with glacial acetic acid (GAA) before human papillomavirus (HPV) testing does not adversely affect HPV detection with the cobas 4800 HPV Test. However, the long-term impact of GAA on HPV DNA was not explored in that study, and inherent cell loss with the lysing protocol used also was observed. The current study considered the long-term effects of GAA treatment of cervical ThinPrep samples on HPV detection with the cobas 4800 HPV Test. They also modified the manufacturer's lysing procedure for ThinPrep specimens to help prevent cell loss.

METHODS

Seventy-eight ThinPrep samples, including previously lysed, archived specimens, were split; then, 1 part was treated with GAA according to the manufacturer's protocol or using a modified protocol, and the other part was left untreated. All samples were tested for HPV using the cobas 4800 HPV Test. The HPV-positive/HPV-negative status of tested samples was used to determine the level of agreement between treated and untreated fractions. Performance of the modified lysing procedure was assessed relative to the manufacturer's protocol.

RESULTS

Positive HPV status produced 97% agreement between the GAA-treated/untreated fractions. Concordance between the GAA-treated/untreated fractions did not differ between the 2 lysing methods; however, the average percentage increases in β-globin and HPV cycle threshold values after lysing were less discernible for the modified lysing method.

CONCLUSIONS

GAA treatment of cervical ThinPrep specimens does not have long-term adverse effects on HPV detection with the cobas 4800 HPV Test. The manufacturer's GAA lysing procedure for ThinPrep specimens can be reliably modified for specimens that may subsequently undergo HPV testing. Cancer (Cancer Cytopathol) 2014;122:250–256. © 2013 American Cancer Society.

INTRODUCTION

In the United Kingdom National Health Service Cervical Screening Program (NHSCSP), the conversion from conventional cytology to liquid-based cytology (LBC), which was completed by 2008, has led to a substantial reduction in the cervical cytology unsatisfactory rate.[1-4] Improvement in sample adequacy with LBC is attributed to the removal of obscuring elements from cervical specimens and the production of a thin cell layer on the slide, both of which enhance the clarity and quality of cervical samples. LBC also permits the treatment of unsatisfactory bloody cervical specimens with an acid wash to lyse blood cells. Cytology laboratories in the United Kingdom routinely reprocess unsatisfactory ThinPrep (Hologic, Bedford, Mass) specimens with 10% glacial acetic acid (GAA) to remove excess blood, thereby restoring satisfactory status.

LBC has also facilitated molecular testing for human papillomavirus (HPV) from the same cervical cytology specimen. The higher sensitivity of HPV testing compared with cytology and the high negative predictive value of the test[5-7] have led to the introduction of HPV testing for triage of borderline cytology and mild dyskaryosis cytology results (NHSCSP terminology)[8] and as a test of cure after treatment for cervical intraepithelial neoplasia (CIN) into routine cervical screening practice in many parts of the United Kingdom.

The equivalent Bethesda terminology[9] for NHSCSP cytology grades[8] of borderline cytology and mild dyskaryosis (low grade) are atypical squamous cells of undetermined significance (ASCUS) and low-grade squamous intraepithelial lesion (LSIL), respectively; the equivalent terminology for moderate and severe dyskaryosis (high-grade dyskaryosis) is high-grade squamous intraepithelial lesion (HSIL); and the equivalent terminology for negative is negative for intraepithelial lesion or malignancy (NILM).

We recently reported on the effect of GAA treatment of cervical ThinPrep specimens on HPV detection with the cobas 4800 HPV Test (Roche Molecular Systems, Pleasanton, Calif).[10] That study clearly indicated that lysing samples with GAA immediately before HPV testing does not adversely affect HPV DNA detection, as demonstrated by the 97.5% agreement on HPV positivity between GAA-treated and untreated samples (kappa = 0.95). However, the study did not consider the long-term impact of GAA treatment on HPV detection with the cobas 4800 HPV Test. This is an important consideration for laboratories that may rely on retrospective HPV testing of stored samples.

We also conducted a UK-wide survey to establish how cytology laboratories are dealing with unsatisfactory LBC specimens that may require an HPV test since the introduction of HPV triage and test of cure into the NHSCSP (unpublished data). Laboratories using the cobas 4800 HPV Test either removed a sample aliquot before lysing (31%) or reported the lysed sample as unsatisfactory for HPV testing (45%). In some instances, laboratories HPV tested the lysed sample (24%). It is not known whether these laboratories conducted their own validation of this process or whether it was assumed that processing with GAA would not have an impact on subsequent molecular tests. Overall, there does not appear to be consensus within the cervical screening community on how to deal with unsatisfactory cervical specimens after the implementation of HPV testing strategies in the United Kingdom. We and others have previously demonstrated that cell loss during GAA processing is an inherent feature[10, 11] and probably can be attributed to Hologic's GAA lysing protocol used in those studies, which involves discarding the supernatant from the first centrifugation of the cervical specimen and replacing it with fresh PreservCyt (Hologic) after lysing of the cell pellet.[12] The sole purpose of this method is to improve cytology preparations; thus, no consideration is given to ancillary molecular testing of the lysed samples. We therefore suggest that modification of the manufacturer's recommended GAA lysing procedure to retain the original supernatant for resuspension of the lysed cell pellet will help to safeguard sample integrity for further molecular testing of cervical specimens.

In the current study, we considered the long-term effects of GAA treatment of cervical ThinPrep samples on HPV detection with the cobas 4800 HPV Test. Several UK cytology laboratories, including our own, use this HPV testing system for triage and test of cure of cervical screening specimens. We also endeavored to validate a modified version of Hologic's recommended GAA lysing procedure for ThinPrep specimens for subsequent HPV testing.

MATERIALS AND METHODS

The study population included cervical screening samples received at Altnagelvin Cytology Department, Western Health and Social Care Trust area in Northern Ireland. The cytology laboratory at Altnagelvin processes about 30,000 cervical ThinPrep samples annually. Currently, 7% to 8% of all cervical specimens received require reprocessing with GAA to yield satisfactory status. Table 1 provides the distribution of the included samples, which were graded according to NHSCSP terminology.[8] After reporting, 61 ThinPrep samples that had not been previously lysed were randomly selected from the HPV triage pool (37 borderline cytology [ASCUS] and 24 mild dyskaryosis [LSIL]) to enrich the study population for HPV positivity. An additional 17 archived specimens (2 severe dyskaryosis and 1 moderate dyskaryosis [HSIL], 2 mild dyskaryosis [LSIL], 9 borderline cytology [ASCUS], and 3 negative [NILM]), which had been lysed with GAA up to 6 months previously, also were included to explore the long-term effects of GAA treatment at different time points. Archived samples had aliquots removed previously before lysing according to the cytology laboratory protocol for the provision of material for ancillary HPV testing. The original cobas 4800 HPV Test results for all triage and archived samples were noted before anonymizing and provided a control for baseline degradation of HPV DNA with time (ie, degradation not associated with exposure to GAA). The total sample selection (n = 78) included nonbloody and bloody specimens. The blood content of the bloody specimens ranged from 0% to 2% and was assessed by reference to the Roche cobas HPV Test Sample Color Reference Guide.[13]

Table 1. Sample Selection for Analysis of the Long-Term Effect of Glacial Acetic Acid Treatment of ThinPrep Specimens on cobas 4800 HPV Test Positivity
 No. of Samples
Cytology ResultaNo.BloodyNonbloodyLyse Method A (Bloody)Lyse Method B (Bloody)Previously Lysed (Bloody)
  1. Abbreviations: HPV, human papillomavirus.

  2. a

    The equivalent Bethesda terminology[9] for National Health Service Cervical Screening Program cytology grades[8] of borderline cytology and mild dyskaryosis (low grade) are atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesion, respectively; and the equivalent terminology for moderate and severe dyskaryosis (high-grade dyskaryosis) is high-grade squamous intraepithelial lesion.

HPV triage pool, n = 61      
Mild2451917(4)7 (1)
Borderline37112614(5)23 (6)
Archived samples, n = 17      
Severe2202 (2)
Moderate1101 (1)
Mild2202 (2)
Borderline9909 (9)
Negative3303 (3)
Total samples78334531 (9)30 (7)17 (17)

The residual ThinPrep specimens from the HPV triage pool were divided into 2 groups to compare the original and modified lysing methods. In the first group, specimens were equally split into 2 fractions: 1 fraction was treated with 10% GAA in CytoLyt (Hologic) according to the manufacturer's recommended protocol[12] (method A; n = 31), and the other fraction was left untreated. In the second group, specimens also were split into 2 equal fractions, with 1 part treated using a modified version of the protocol (method B; n = 30) and the other part left untreated. Briefly, treated specimens were centrifuged at ×1200g for 5 minutes, and the PreservCyt (Hologic) supernatant was either discarded (method A) or poured back into the original ThinPrep vial (method B) before the addition of 30 mL of the GAA/CytoLyt solution to each cell pellet. The specimens were centrifuged again at ×1200g for 5 minutes, and each pellet was resuspended in fresh PreservCyt to the initial volume of the split sample (method A) or in the appropriate retained PreservCyt supernatant (method B). The archived samples were lysed previously according to the manufacturer's recommended protocol (method A). The untreated samples (archived aliquots and triage split samples) and treated samples (GAA-lysed archived and triage split samples) were tested for HPV using the cobas 4800 HPV Test, which simultaneously detects a total of 14 high-risk HPV (hrHPV) genotypes, including 12 pooled types (HPV31, HPV33, HPV35, HPV39, HPV45, HPV51, HPV52, HPV56, HPV58, HPV59, HPV66, and HPV68) and 2 individually identified types (HPV16 and HPV18). The cobas 4800 HPV Test also includes a separate β-globin control to ensure specimen adequacy. Strict assay and decontamination procedures were followed to avoid false-positive results because of specimen contamination, as described previously.[10] Separate, dedicated areas and equipment were used for sample preparation, reagent preparation, and polymerase chain reaction analysis. The cobas 4800 HPV Test kit also contains a proprietary AmpErase enzyme to avoid carryover of polymerase chain reaction product. The HPV-positive/negative status of tested samples determined the level of agreement between treated and untreated fractions. Statistical analysis was performed using an unweighted kappa. The performance of the modified GAA lysing procedure was assessed relative to the manufacturer's recommended protocol and based on the level of agreement on HPV positivity between GAA-treated and untreated fractions and variance in the degree of increase in cycle threshold (Ct) values after lysing between the 2 methods. Microscopic analysis of ThinPrep slides prepared from samples that were treated using both the manufacturer's recommended method and the modified GAA lysing method was performed by 1 biomedical scientist and 1 cytopathologist to ensure that the quality of the LBC slides prepared using the latter method was not compromised.

RESULTS

Valid results were obtained for all 78 sample pairs. An analysis of overall HPV positivity indicated 97% agreement (76 of 78 sample pairs) between the GAA-treated and untreated fractions (kappa = 0.94). There were 49 of 51 (96.1%) concordant HPV-positive results and 27 of 27 (100%) concordant HPV-negative results (Table 2). The 2 discordant results, which occurred in triage samples (HPV tested at 24 days postlysing), were initially HPV-positive but subsequently tested negative after GAA treatment (Table 2). One sample was positive for HPV18 (Ct = 40.0), and the other sample was positive for 12 other hrHPV types (Ct = 37.4). There were no discrepancies in archived samples (HPV tested between 29 and 201 days postlysing), which indicated complete agreement on HPV positivity between treated and untreated fractions at all time points (Table 2). The current HPV results for untreated samples demonstrated absolute agreement on HPV positivity with the original HPV results for all untreated triage and archived samples. The degree of concordance between the GAA-treated and untreated fractions indicated that there was no difference between the 2 lysing methods (Fig. 1).

Table 2. Long-Term Effect of Glacial Acetic Acid Treatment of ThinPrep Specimens on cobas 4800 HPV Test Positivity
   No. of Samples
  1. Abbreviations: GAA, glacial acetic acid; HPV, human papillomavirus; HPV-Neg, HPV-negative; HPV-Pos, HPV16-positive and/or HPV18-positive, and/or positive for 12 other high-risk HPV types.

  2. a

    HPV results in untreated samples represent both the current results for the untreated fractions and the original HPV results for the same samples, because there was absolute agreement on HPV-positive status upon retesting. Agreement between the treated and untreated fractions was 97% (kappa value = 0.94).

  3. b

    The equivalent Bethesda terminology[9] for National Health Service Cervical Screening Program cytology grades[8] of borderline cytology and mild dyskaryosis (low grade) are atypical squamous cells of undetermined significance and low-grade squamous intraepithelial lesion, respectively; and the equivalent terminology for moderate and severe dyskaryosis (high-grade dyskaryosis) is high-grade squamous intraepithelial lesion.

   Untreated SamplesaGAA-Treated Samples
cobas 4800 HPV Test Cytology ResultbNo.Time to HPV Test Post-Treatment, dHPV-PosHPV-NegHPV-PosHPV-Neg
HPV triage pool, n = 61      
Mild2424222222
Borderline372418191621
Archived samples, n = 17      
Severe229-802020
Moderate1291010
Mild2178-2011111
Borderline982-2017272
Negative378-800303
Total samples78 51274929
Figure 1.

Glacial acetic acid (GAA) lysing methods are compared based on the average percentage increase in β-globin cycle threshold (Ct) values and the percentage agreement on positive human papillomavirus (HPV) status between GAA-treated/untreated fractions.

An analysis of Ct values was performed on the current HPV results. β-Globin Ct values were available for 76 of 78 sample pairs, and HPV Ct values were available for 48 of 49 concordant HPV-positive sample pairs. GAA-treated fractions had higher β-globin Ct values in 66 of 76 sample pairs (86.84%) and higher HPV Ct values in 38 of 48 sample pairs (79.17%) compared with untreated fractions. The average percentage increase in β-globin Ct values postlysing was less discernible for lysing method B (7.87%) compared with lysing method A (9.72%) (Fig. 1). Similarly, the average percentage increase in Ct values for the 12 other hrHPV and for HPV16 after lysing was less evident for lysing method B (13.38% and 8.16%, respectively) compared with lysing method A (14.12% and 11.2%, respectively).

A microscopic comparison of ThinPrep slides prepared from samples lysed with GAA using the manufacturer's recommended protocol and the modified lysing method did not reveal any evidence of compromised quality in slides prepared from the latter method. The clarity of all slides was considered standard or better for GAA-treated samples, and all slides were regarded as satisfactory for reporting.

DISCUSSION

In the United Kingdom, cytology laboratories routinely lyse unsatisfactory cervical LBC specimens with GAA to facilitate cytologic analysis and reduce the cervical cytology unsatisfactory rate. We conducted a UK-wide survey to establish how cytology laboratories are dealing with unsatisfactory LBC specimens that may require an HPV test since the introduction of HPV triage and test of cure into the NHSCSP and observed disparity within the cervical screening community on approaches to deal with these specimens (unpublished data from the author's laboratory).

We recently published data on the effect of GAA treatment of cervical ThinPrep samples on HPV detection with the cobas 4800 HPV Test.[10] That study demonstrated a high degree of concordance between GAA-lysed and untreated aliquots, with overall agreement on HPV positivity of 97.5% (117 of 120 samples) between treated and untreated fractions (kappa = 0.95). Ct values for β-globin controls and HPV, which are inversely proportional to the amount of target DNA in samples, generally were higher in treated aliquots, indicating loss of cellular material during GAA processing. However, in that study, we did not consider the long-term impact of GAA treatment on HPV detection with the cobas 4800 HPV Test.

It has been suggested that low pH could affect the molecular detection of HPV in LBC samples.[14] The findings of our original GAA study challenge this theory10; however, the proposed detrimental effect on DNA required consideration of the effects of prolonged exposure to acidic conditions. We reported previously that the average pH of specimens during GAA treatment is 2.95; after treatment, specimen pH is immediately increased to an average of 4.82 (unpublished data from the author's laboratory).[10] Although this environment remains acidic compared with the 5.85 average pH of specimen vials before GAA processing (unpublished data from the author's laboratory), in our original GAA study, all specimens were HPV tested within 24 hours of processing.[10] Conversely, the current study included specimens that had been lysed with GAA up to approximately 6 months before HPV testing. This represented the recommended maximum time limit for storage of PreservCyt samples to sustain reliable HPV detection with the cobas 4800 HPV Test.[15]

The results of the current study clearly demonstrate that GAA treatment of ThinPrep samples does not have significant long-term effects on HPV detection with this system. Overall, HPV positivity had 97% (76 of 78 samples) agreement (96.1% concordant HPV-positive results and 100% concordant HPV-negative results) between the treated and untreated fractions. The absolute agreement on HPV positivity between the original and current HPV results from untreated samples indicated that HPV DNA had not naturally degraded with time; therefore, the concordance on HPV positivity between the current GAA-treated and untreated fractions can be confidently attributed to the sustained stability of HPV DNA after GAA treatment of specimens. There were 2 discordant results in the triage pool (HPV tested at 24 days postlysing); however, archived samples demonstrated complete agreement on HPV positivity between treated and untreated fractions at all time points (HPV tested between 29 and 201 days postlysing), indicating that DNA stability and concomitant HPV detection were not compromised even at the limit of reliable HPV detection with the cobas 4800 HPV Test.[15]

Comparable to our earlier findings, 1 of the discordant cases, which was initially HPV-positive but subsequently tested negative after GAA-treatment using the original lysing method,[12] had an initial HPV18 Ct value of 40. According to the package insert for the cobas 4800 HPV Test,[15] this appears to be the limit of detection for the assay; thus, HPV results close to this threshold could waiver between positive and negative. The other discordant case was initially positive for 12 other hrHPV types (Ct = 37.4) but subsequently tested negative postlysing with the modified protocol. Whereas the Ct value relating to this discrepancy was lower than in the previous case, the Ct values for the β-globin control of this sample were 28.1 and 30.4 in the untreated and treated fractions, respectively.

Overall, the higher Ct values for β-globin and HPV associated with GAA treatment of samples in the current study (86.84% and 79.17% of cases, respectively) also are comparable to previous findings[10, 11] and further support the theory that cell loss during GAA processing is an inherent feature and can probably be attributed to the GAA lysing protocol used in those studies. Therefore, we sought to improve the manufacturer's recommended GAA lysing procedure, which involves discarding the supernatant from the first centrifugation of the cervical specimen and replacing it with fresh PreservCyt after lysing of the cell pellet.[12] We modified the protocol to retain the original supernatant to resuspend the lysed cell pellet, thereby optimizing the maintenance of cell concentration for subsequent HPV testing. The degree of concordance between GAA-treated and untreated fractions did not indicate any difference between the 2 lysing methods; however, there was notable variation in the level of increase in Ct values postlysing between the 2 methods. The average percentage increase in β-globin Ct values postlysing was less discernible for the modified lysing method B (7.87%) compared with the original lysing method A (9.72%) (Fig. 1). Similarly, the average percentage increase in Ct values for the 12 other hrHPV types and for HPV16 after lysing also was less evident with the modified lysing method (13.38% and 8.16%, respectively) compared with the original method (14.12% and 11.2%, respectively).

The original GAA-lysing protocol, which was optimized by Hologic to improve cytology preparations, does not consider ancillary molecular tests and has the potential for false-negative HPV results caused by loss of epithelial cells in the discarded supernatant or alteration of cell concentration if the initial sample volume is exceeded when resuspending lysate in fresh PreservCyt. Hologic's protocol states that lysates should be filled to a volume of 20 mL with fresh PreservCyt; however, the initial volume of many samples is less than 20 mL before lysing. Therefore, strict adherence to this will result in dilution of samples. In the current study, retaining the original PreservCyt supernatant after the first centrifugation, instead of using fresh PreservCyt after lysing, almost certainly minimized cell loss and ensured that samples were not overfilled. This was reflected in the lower β-globin and HPV Ct values obtained for specimens that were lysed using this modified method. Notably, the risk of cell loss when decanting the GAA lysing solution after the second centrifugation cannot be discounted. Although the cell pellet at this stage is much tighter than the “fluffy” pellet that results from the first centrifugation step and is less prone to loss of cellular material, the conversion to HPV-negative status and the associated increase in β-globin Ct value postlysing observed in 1 sample that was lysed using the modified method may be indicative of this.

The current results provide assurance of the long-term stability of HPV DNA in ThinPrep samples after GAA lysing. This is an important consideration in the management of unresolved colposcopy and cytology/histology mismatch cases discussed at multidisciplinary team meetings, which may rely on retrospective HPV testing of stored samples. A recent study investigating the utility of HPV testing to resolve uncertainty in colposcopy indicated that, of the 5 categories of clinical problems identified, HPV testing could help to streamline management in all scenarios.[16]

Validation of the modified version of Hologic's GAA lysing procedure for ThinPrep specimens has also been provided by the level of agreement on HPV positivity after lysing between the 2 methods. Furthermore, cytologic analysis of the bloody specimens did not reveal any difference in clarity between samples lysed with either method, providing assurance that cytology preparations will not be adversely affected if the modified GAA lysing procedure is used.

If cytology laboratories decide to HPV test GAA-lysed samples, then they need to safeguard the reliability of cervical samples and take measures to reduce the risk of contamination, as any handling or processing of samples before molecular testing could jeopardize sample integrity. We used strict decontamination procedures and control measures throughout to reduce the risk of contaminating specimens. Processing samples using the modified GAA-lysing protocol also helps to preserve sample integrity, because reuse of the original PreservCyt eliminates the risk of contamination presented by PreservCyt pump dispensers when refilling samples with fresh PreservCyt.

In conclusion, this study has demonstrated that GAA treatment of cervical ThinPrep specimens does not have long-term adverse effects on HPV detection with the cobas 4800 HPV Test. The study has also provided validation of the reliability of a modified version of Hologic's GAA lysing procedure for ThinPrep specimens that may subsequently undergo HPV testing.

FUNDING SUPPORT

No specific funding was disclosed.

CONFLICT OF INTEREST DISCLOSURES

Drs. McMenamin and McKenna have both received nonfinancial support in the form of cobas 4800 HPV test kits from Roche Molecular Diagnostics for a previous study.

Ancillary