Abnormal Cervicovaginal Cytology With Negative Human Papillomavirus Testing

280 Abnormal Cervicovaginal Cytology With Negative Human Papillomavirus Testing Giovanni Negri, MD Bettina Rigo, BS Fabio Vittadello, ScD Christine Mian, ScD Eduard Egarter-Vigl, MD Department of Pathology, Central Hospital Bolzano, Bolzano, Italy. BACKGROUND. Human papillomavirus (HPV) testing has become part of cervicovaginal diagnostics in many laboratories. False-negative HPV results, even if rare, are potentially relevant clinically, particularly when testing is used alone. In this study, the authors investigated the occurrence and causes of abnormal cervicovaginal cytologies with negative HPV testing. METHODS. The study was performed on 4130 liquid-based cytology (LBC) specimens from 2918 women who had abnormal cervical cytology or colposcopy or a history of abnormal cervical smear. HPV testing with Hybrid Capture II was repeated in LBC specimens with obviously atypical cytology (squamous intraepithelial lesion or abnormal squamous cells, cannot exclude high-grade lesion) and initial negative test results. The results were compared with the follow-up results and with the cytologic features of the specimens. RESULTS. Of 4130 LBC specimens, 94 (2.28%) showed obviously abnormal cells despite negative HPV results, and 50 of those 94 samples (53.19%) were HPVpositive on a repeated testing of the same specimen. Histologic follow-up showed high-grade intraepithelial lesion or carcinoma in 10 specimens that initially were negative for HPV. Eight of those samples were HPV-positive on testing repetition. Both specimens that had persistently negative results had poor-quality cytologic samples at revision. CONCLUSIONS. High-grade lesions may be identified in the follow-up of HPV-negative specimens with abnormal cytology. Testing repetition in patients with discordant cytology may prevent most potentially clinically relevant false-negative HPV results. Cancer (Cancer Cytopathol) 2007;111:280 4. Ó 2007 American Cancer Society. KEYWORDS: human papillomavirus, Hybrid Capture II, liquid-based cytology, ThinPrep. We thank Ventana Italy for supplying the human papillomavirus testing kit free of charge and Ms. Alessandra Spada and Mr. Hubert Dorfmann for helping collect the cases. Address for reprints: Giovanni Negri, MD, Department of Pathology, Central Hospital Bolzano, Via Boehler 5, 39100 Bolzano, Italy; Fax: (011) 00390471908175; E-mail: ginegri@gmail.com Received February 22, 2007; revision received May 19, 2007; accepted May 22, 2007. Human papillomavirus (HPV) testing has become part of routine cervicovaginal diagnostics in many laboratories. The Atypical Squamous Cells of Undetermined Significance/Low-Grade Squamous Intraepithelial Lesions Triage Study 1 demonstrated the utility of HPV testing for the triage of women with abnormal squamous cells (ASC), and the possibility of replacing cytology-based, primary screening with molecular techniques has been suggested by several authors. 2 5 Although the scenario of cervical cancer prevention is changing because of the introduction of HPV vaccines, HPV testing still will have several applications, including the follow-up of women who already have undergone vaccination. Among testing techniques, Hybrid Capture II (HCII) (Digene, Gaithersburg, Md) is one of the most widespread because of its simplicity and high sensitivity. When combined with liquid-based ª 2007 American Cancer Society DOI 10.1002/cncr.22952 Published online 27 August 2007 in Wiley InterScience (www.interscience.wiley.com).

CV Cytology With Discordant HPV Testing/Negri et al. 281 cytology (LBC), HPV testing with HCII can be useful for the interpretation of troublesome cases without the need for recalling the woman. 1,6 With LBC, only a part of the sampled cells is used for the cytologic slides, and previous studies have indicated that HPV testing with HCII on the residual cell suspension has high sensitivity comparable to that of polymerase chain reaction (PCR) 7 9 and high interlaboratory reproducibility. 10,11 However, false-negative HPV testing can occur. In a previous study, we reported that false-negative HPV results with HCII on LBC specimens may be caused by insufficient cellularity of the specimen. 12 One other recent study indicated that false-negative HCII results may occur because of a low HPV DNA load. 13 Zuna et al. 14 observed in their study that from 3% to 11% of low-grade squamous intraepithelial lesions (LSILs) were negative for HPV but that from 12% to 32% of those lesions turned HPV-positive at the successive follow-up test. In that study, however, subsequent testing was performed on different samples. Taking into consideration the possibility of false-negative results becomes important in cases of obviously abnormal cytology and negative HPV tests. Even if they are rare, those apparently false-negative results may have important clinical consequences and may cause distrust in molecular techniques. In the current study, HPV testing with HCII was repeated in 94 ThinPrep specimens that had abnormal cytology immediately after an initial negative test. Results were compared with follow-up results, and cytologies were analyzed in an attempt to clarify the cause of the discordance. MATERIALS AND METHODS During the period from 2002 to 2004, HPV testing with HCII was performed on residual cell suspensions of 4130 ThinPrep specimens from 2918 women. Testing was routinely performed because of the presence of atypical squamous cells (ASC) in the LBC or was requested by the woman s gynecologist because of abnormal colposcopy or a history of ASC or LSIL. For HCII testing, 6 ml of the cell suspension were used. After centrifugation, HPV testing with the high-risk HPV probe (HRHPV) (including HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68) was performed according to the manufacturer s recommendations. Test results with relative light units (RLU)/cut-off (RLU/CO) values \1.0 were considered negative. Negative controls, positive calibrators, and quality controls always were included in the tests. In 94 specimens (2.28%), HPV testing was repeated after a first negative test result because of TABLE 1 Results of Repeated Human Papillomavirus Testing on 94 Specimens With Previously Negative Test Results HPV test result No. of test % Positive 50/94 53.19 Negative 44/94 46.81 HPV indicates human papillomavirus. evident discordance with cytology results. Discordant cases were defined by a consensus group as women who had negative HRHPV test results and obviously abnormal epithelial cells, all strongly suggestive of ASC, cannot rule out high-grade SIL (HSIL) or SIL. The second test was performed using the same protocol as the first test. In 90 specimens, however, testing was also performed using the low-risk HPV (LRHPV) probe to exclude a pure low-risk HPV infection. All women had at least cytologic follow-up. Thirteen women also had histologic follow-up. In situ hybridization with the Inform HRHPV probe cocktail (Ventana Medical Systems, Tucson, Ariz; including HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 68, and 70) was performed according to the manufacturer s specifications on histologic specimens of high-grade lesions (cervical intraepithelial neoplasia type 2 or worse [CIN21]) that showed a persistently negative HPV test results with HCII on LBC. Frequencies and percentages were computed for women who had abnormal cytology despite a negative HPV test. The 95% confidence interval (95% CI) of the difference between proportions was constructed using the Newcombe refinement on a Wilson score-based method. 15,16 RESULTS Overall, considering all 4130 LBC specimens, HRHPV testing was positive in 90.9% of LSIL and in 97.8% HSIL. Histologic follow-up was available on 426 women, including 140 women who had negative results, 145 women who had CIN1, and 141 women who had CIN21. In 94 of 4130 LBC specimens (2.28%), HPV testing was repeated after an initial negative test result (RLU/CO \1.0; mean, 0.35) because of an evident discordance with cytology results. In 44 of those 94 retested LBC specimens (46.81%), a negative HRHPV result was confirmed (Table 1); however, in 23 of the 44 specimens (52.27%), LRHPV was positive. The histologic followup of those 44 women showed that 3 women had CIN1, 1 woman had CIN2, and 1 woman had largely invasive squamous cell carcinoma. Cytologically, 6

282 CANCER (CANCER CYTOPATHOLOGY) October 25, 2007 / Volume 111 / Number 5 TABLE 2 Histologic Follow-up and Results From the Second Testing With High-risk Human Papillomavirus Second testing Histology CIN1 CIN2 CIN3 AIS Carcinoma Positive 0 2 5 1 0 Negative 3 1 0 0 1 Total 3 3 5 1 1 CIN indicates cervical intraepithelial neoplasia; AIS, adenocarcinoma in situ. TABLE 3 Cytologic Revision of 10 Cytologies With Initially Negative Human Papillomavirus Results Who Had Histologic Follow-up of a High-grade Lesion Patient no. HPV repetition Follow-up Cytologic revision 1 LR2 HR1 CIN3 Scant cellularity, few 2 LR1 HR1 CIN2 Satisfactory for evaluation, numerous 3 LR1 HR1 CIN3 Satisfactory for evaluation, no 4 LR2 HR1 CIN3 Satisfactory for evaluation, inflammation, numerous 5 LR1 HR1 CIN2 Satisfactory for evaluation, high cellularity, numerous 6 LR1 HR1 AIS Satisfactory for evaluation, numerous 7 LR1 HR1 CIN3 Satisfactory for evaluation, numerous 8 LR2 HR1 CIN3 Scant cellularity, few 9 LR2 HR2 Carcinoma Scant cellularity, hemorrhagic, few 10 LR1 HR2 CIN2 Scant cellularity, few cells with slight atypia HPV indicates human papillomavirus; LR indicates low risk; 2, negative; HR, high risk; 1, positive; CIN, cervical intraepithelial neoplasia; AIS, adenocarcinoma in situ. additional women showed LSIL in follow-up. All other women had negative follow-up. Two of the women with CIN1 and the 1 woman with CIN2 had positive results for LRHPV. In situ hybridization with HRHPV probes on the histologic specimens from the women who had persistently HCII-negative squamous cell carcinoma and CIN2 was positive with a typical, dot-like, nuclear pattern. In 50 of 94 specimens (53.19%) who had a repeat HPV test, HRHPV results changed to positive (Table 2). In 21 specimens (42%), HRHPV was associated with a positive LRHPV result. In 21 samples (42%), the HRHPV-RLU value at the second test was [1000; and, in 31 women (62%), the second test value was [500. There was an RLU/CO ratio [2.5 in 33 specimens (66%). The mean RLU/CO value of the 17 samples who had an RLU/CO value \2.5 was 1.28. In the histologic follow-up of this group, there were 2 women with CIN2, 5 women with CIN3, and an adenocarcinoma in situ. Overall, all but 1 women with CIN21 had an RLU/CO value [2.5 (mean, 488,65 RLU/CO) on a repeat test. One woman with CIN2 had an RLU/CO value of 1.85. Overall, histologic follow-up identified a dysplastic lesion in 13 of 94 initially HPV-negative cases (13.83%), including 10 high-grade lesions (CIN21) (Table 2). Eight of 10 women (80%) with CIN21 had positive HRHPV results at the second testing. The 2 women with high-grade lesions in which negative HRHPV results were was confirmed included 1 woman who had invasive squamous cell carcinoma with few neoplastic cells in a highly hemorrhagic and necrotic background and 1 woman who had CIN2 with cytologically only very few, slightly dysplastic cells that had been classified as LSIL and was positive for LRHPV (Table 3). At the initial testing, 2.28% of LBC results (95% CI, 1.86 2.78%) showed abnormal cytology despite a negative HPV result. After repeated HPV testing, the overall percentage was reduced to 1.07% (95% CI, 0.80 1.43%), with a difference (21.21%) that differed statistically significantly from zero (P \.0001; 95% CI, 21.59% to 20.89%). DISCUSSION The results from this study indicated that HPV testing with HCII on ThinPrep residual cell suspension is mostly reliable but may be afflicted by some falsenegative results, although only a few of those results have potential clinical significance. However, repeat testing in doubtful cases can reduce significantly the number of false-negative results. In fact, of 4130 specimens, 2.28% showed an abnormal cytology despite a negative HPV test; and, after repeated testing, the overall percentage was reduced to 1.07%. Eight of 10 CIN21 cases changed to HPV-positive at the second test, and 7 of those specimens showed an RLU/CO ratio well over the retest zone, which included values between 1 and 2.5. 17 Among the women who had persistently HPVnegative results, only 2 had clinically significant follow-up, and both were characterized by low sample quality and few abnormal cells. The treatment of these women, however, along with the treatment for women who had results that changed to HPV positive at the second testing, would have been delayed

CV Cytology With Discordant HPV Testing/Negri et al. 283 if cytology had not been performed at the same time. Although the cause of the persistently HPV-negative CIN1 results obviously was low sample quality, the cause of the first negative HPV result in women who had results that changed to positive at the second testing is more difficult to comprehend. Repeating HPV testing on the same sample may be associated with a reduction of the RLU value, which can lead in some specimens to a false-negative test result. 12 Because most specimens had a clearly positive RLU/CO ratio at the second testing, we conclude that the original viral load already should have been high enough to warrant a clear positive result at the first testing. Furthermore, these cases always were sporadic, and testing was performed together with correctly positive and negative routine cases as well as positive and negative controls. Thus, the most probable explanation for these cases is sporadic human error in performing the test. Because 3 different technicians were involved in HPV testing during the study period, and the false-negative results were distributed uniformly over the time, it is difficult to individuate a part of the manual procedure involving a higher risk of error. In the future, however, it is expected that the introduction of complete automation of the procedure may reduce at least in part the occurrence of such false-negative test results. Because, in the current study, only a selected high-risk population was analyzed, it is difficult to draw conclusions that may apply to a screening setting. Potentially clinically relevant false-negative HPV test results are rare, and the risk of missing highgrade lesions is relatively low. Overall, the percentage of specimens who had CIN21, HRHPV-negative results at the first testing was 0.24%. Because, during the same period, 553 CIN21 lesions were identified at our institution, this would include 1.81% of all clinically relevant lesions. Furthermore, clinical follow-up essentially is driven by cytology in women who have abnormal results. However, from a practical point of view, HPV testing repetition may prevent the misinterpretation as reactive of some women who have an ASC diagnosis with a false-negative HPV test. If cytologists know the HPV test results at the time of cytologic diagnosis, which is the case in many laboratories, a false-negative test result may lead to an underestimation of the lesion and delayed colposcopy. The awareness of possible sources of false-negative HPV tests is important particularly in the context of HPV only-based screening. 18 Our data alone may not support keeping the Papanicolaou test alongside HPV testing for primary screening. However, at least in high-risk settings or in follow-up, HPV testing on LBC specimens has the advantage of allowing an immediate repetition of the test in case of discordant results. Our results indicate that this approach can clarify most false-negative results, also allowing the identification of those tests with inadequate cytologic sampling that should be repeated immediately. REFERENCES 1. Schiffman M, Solomon D. Findings to date from the ASCUS-LSIL Triage Study (ALTS). Arch Pathol Lab Med. 2003;127:946 949. 2. Ronco G, Giorgi-Rossi P, Carozzi F, et al. Human papillomavirus testing and liquid-based cytology in primary screening of women younger than 35 years: results at recruitment for a randomised controlled trial. Lancet Oncol. 2006;7:547 555. 3. Goldie SJ, Kim JJ, Wright TC. Cost-effectiveness of human papillomavirus DNA testing for cervical cancer screening in women aged 30 years or more. Obstet Gynecol. 2004;103:619 631. 4. Kulasingam SL, Hughes JP, Kiviat N, et al. Evaluation of human papillomavirus testing in primary screening for cervical abnormalities: comparison of sensitivity, specificity, and frequency of referral. JAMA. 2002;288:1749 1757. 5. Schiffman M, Herrero R, Hildesheim A, et al. HPV DNA testing in cervical cancer screening: results from women in a high-risk province of Costa Rica. JAMA. 2000;283:87 93. 6. Sarode VR, Werner C, Gander R, et al. Reflex human papillomavirus DNA testing on residual liquid-based (TPPT) cervical samples: focus on age-stratified clinical performance. Cancer (Cancer Cytopathol). 2003;99:149 155. 7. Riethmuller D, Gay C, Bertrand X, et al. Genital human papillomavirus infection among women recruited for routine cervical cancer screening or for colposcopy determined by Hybrid Capture II and polymerase chain reaction. Diagn Mol Pathol. 1999;8:157 164. 8. Cope JU, Hildesheim A, Schiffman MH, et al. Comparison of the Hybrid Capture 2 test and PCR for detection of human papillomavirus DNA in cervical specimens. J Clin Microbiol. 1997;35:2262 2265. 9. Clavel C, Masure M, Putaud I, et al. Hybrid Capture II, a new sensitive test for human papillomavirus detection. Comparison with Hybrid Capture I and PCR results in cervical lesions. J Clin Pathol. 1998;51:737 740. 10. Carozzi FM, Del Mistro A, Confortini M, et al. Reproducibility of HPV DNA testing by Hybrid Capture 2 in a screening setting. Am J Clin Pathol. 2005;124:716 721. 11. Castle PE, Wheeler CM, Solomon D, et al. Interlaboratory reliability of Hybrid Capture 2. Am J Clin Pathol. 2004;122: 238 245. 12. Negri G, Rigo B, Vittadello F, et al. Human papillomavirus typing with Hybrid Capture II on archived liquid-based cytologic specimens: is HPV typing always reproducible? Am J Clin Pathol. 2004;122:90 93. 13. Jastania R, Geddie WR, Chapman W, Boerner S. Characteristics of apparently false-negative Digene Hybrid Capture 2 high-risk HPV DNA testing. Am J Clin Pathol. 2006;125: 223 228. 14. Zuna RE, Wang SS, Rosenthal DL, et al. Determinants of human papillomavirus-negative, low-grade squamous intraepithelial lesions in the Atypical Squamous Cells of Undetermined Significance/Low-Grade Squamous Intraepithelial

284 CANCER (CANCER CYTOPATHOLOGY) October 25, 2007 / Volume 111 / Number 5 Lesions Triage Study (ALTS). Cancer (Cancer Cytopathol). 2005;105:253 262. 15. Newcombe R. Improved confidence intervals for the difference between binomial proportions based on paired data. Stat Med. 1998;17:2635 2650. 16. Newcombe RG, Altman DG. Proporzioni e loro Differenze. In: Altman DG, Machin D, Bryant TN, et al, eds. Statistica Medica. 2nd ed. Torino, Italy: Minerva Medica; 2004:52 56. 17. Muldrew KL, Beqaji SH, Han J, et al. Evaluation of a Digene-recommended algorithm for human papillomavirus low-positive results present in a retest zone. Am J Clin Pathol. 2007;127:97 102. 18. Stoler MH, Castle PE, Solomon D, Shiffman, M. The expanded use of HPV testing in gynecologic practice per ASCCP-guided management requires the use of wellvalidated assays. Am J Clin Pathol. 2007;127:335 337.