Abstract
The Hybrid Capture 2 (HC2) test targets 13 human papillomavirus (HPV) types. Here, cross-reactivity with non-HC2-targeted HPV types is described. We aimed to define the proportion of HC2-positive women who had negative results with HC2-targeted HPV types and estimate its determinants and impact on women's health management. The New Technologies for Cervical Cancer (NTCC) trial was followed in two predetermined phases. Women in the experimental arm were tested for the presence of HPV DNA by HC2 following a sample collection in PreservCyt (first phase) or Digene specimen transport medium (STM) (second phase). HPV genotyping was performed on DNA samples from HC2-positive women by PCR with GP5+/GP6+ primers and reverse line blot (RLB) hybridization. Untyped samples were submitted to direct sequencing or restriction fragment length polymorphism. Multivariate logistic regression analysis estimated the adjusted odds ratios (ORs) between the presence of HC2-targeted types and age, viral load, and type of transport medium. Out of 2,920 HC2-positive samples, 2,310 (79.1%) were positive on RLB for HC2-targeted types, 396 were positive (13.6%) for only non-HC2-targeted types (mostly represented by HPV-53, HPV-66, and HPV-70), and in 214 (7.33%) samples, no HPV types were detected. The probability of detecting HC2-targeted types increased with increasing viral load expressed as the relative light unit/positive-control specimen ratio (RLU/PC) (OR for unitary increase of log RLU/PC, 1.35; 95% confidence interval [CI], 1.30 to 1.42) and with STM versus PreservCyt (OR, 1.56; 95% CI, 1.25 to 1.84). If only the samples containing HC2-targeted types tested positive, the positive predictive value (PPV) would have increased from 7.0% (95% CI, 6.1% to 8.0%) to 8.4% (95% CI, 7.3 to 9.6), although 4.9% (95% CI, 2.4% to 8.8%) of cervical intraepithelial neoplasia grade 2+ (CIN2+) cases would have been missed. In conclusion, STM use and an increased cutoff would reduce the HC2 analytical false-positive rate and increase the positive predictive value for high-grade CIN. The gain in clinical sensitivity by detecting non-HC2-targeted HPV types is limited.
INTRODUCTION
Cervical screening has the main purpose of decreasing the burden of cervical cancer by detecting and treating high-grade cervical intraepithelial neoplasia (CIN). Highly sensitive and specific tests have been established to identify the human papillomavirus (HPV) infections that are associated with detectable CIN. Double-testing studies (1) and randomized controlled trials (RCTs) (2–6) have highlighted Hybrid Capture 2 (HC2) as a highly sensitive test (>95%) (7) for detecting high-grade CIN. Conversely, the HC2 test showed lower clinical specificity than did cytology (90%) (7). Most of the false positives that make the test specificity so low are due to women who are actually infected by a high-risk HPV type but who have not developed high-grade lesions and, in most cases, will never develop cervical lesions (8). To date, this drawback has been overcome by the introduction of triage procedures that limit the referrals of HPV-positive women for unnecessary diagnostic procedures.
It has been claimed that HC2 also has analytical false-positive results. HC2 includes a cocktail of probes designed to detect 13 HPV types, HPV-16, -18, -31, -33, -35, -39, -45, -51, -52, -56, -58, and -59, which are classified by the International Agency for Research on Cancer (IARC) (9) as carcinogenic with sufficient evidence (group 1), and HPV-68, which is classified as probably carcinogenic (group 2A). Some studies have shown cross-reactivity with other HPV types that are phylogenetically related to the targeted genotypes (10–19).
Increasing analytical false positives were observed with decreasing viral load, as measured by the ratio between the relative light units (RLU) of the specimen and the RLU of a positive cutoff (PC) consisting of 1 pg/ml of HPV DNA (RLU/PC ratio) (11, 15, 20–22). In addition, lower reproducibility of the HC2 results between laboratories (23) and lower agreement between HC2 and PCR with the MY09 to MY11 primers when followed by dot blot hybridization (19) was reported with samples collected in PreservCyt (which is used for liquid-based cytology) than in those collected in the Digene standard transport medium (STM); this suggests a role of the transport medium in the analytical accuracy of HC2. The large size of the New Technologies for Cervical Cancer (NTCC) trial, in which samples were collected in both the PreservCyt and STM transport media and were stored, and the active follow-up of the women involved, provide a suitable frame for accurate estimates of these parameters.
The objective of this study was to estimate the frequency of analytical false-positive results of HC2, its determinants, and their impacts on clinical accuracy.
MATERIALS AND METHODS
Study population and sample collection.
The New Technologies for Cervical Cancer (NTCC) screening study was a randomized controlled trial conducted in 9 Italian centers that compared cytology to an experimental arm during two subsequent phases, the first with cotesting for HPV DNA by the Hybrid Capture 2 (HC2) (Digene-Qiagen, Hilden, Germany) test and liquid-based cytology and the second with stand-alone HC2 testing (2, 4, 24). The study was approved by the local ethics committees of the participating centers. The randomized clinical trial registration number is ISRCTN81678807. Women were considered to be HC2 positive if their RLU/PC ratio was >1, the cutoff that was recommended by the manufacturer. Seventy-four percent of eligible women were accepted for randomization in the trial, and 47,369 were randomized to the experimental arm.
The storage of residual cell samples after a positive HC2 test was set up during both study phases in Florence, Padua, Trento, and Turin, while in Bologna, Imola, Ravenna, and Viterbo, it was only performed during the second phase, and it was not done at all in Verona. Of 37,367 women randomized to the experimental arm in these time periods and centers, 37,077 had valid HC2 results and 3,033 of them were found to be HPV positive. Only the first HC2-positive sample for each woman was genotyped. Samples were not available for 113 HPV-positive women, leaving 2,920 samples for genotyping. Among them, 1,264 (43%) were originally collected in PreservCyt (Cytyc Corporation, Marlborough, MA) and processed with a manual conversion procedure (HC2 sample conversion kit; Digene-Qiagen, Hilden, Germany) before performing the HC2 test during phase 1, and 1,656 (57%) were collected in the Digene STM (Digene-Qiagen, Hilden, Germany) solution during phase 2. Samples from Bologna, Imola, Ravenna, and Viterbo represented 25.7% (425/1,656) of those obtained in phase 2. Residual cells from samples collected in PreservCyt were stored at −80°C following double phosphate-buffered saline (PBS) washing to remove methanol; aliquots (400 μl) from samples collected in STM were taken before HC2 testing and stored at −80°C in their own solution.
DNA extraction.
DNA was obtained from one of the two stored aliquots of PreservCyt and STM samples by purification through the QIAamp DNA minikit (Qiagen, Hilden, Germany) according to the manufacturer's instructions, with a final elution in 80 to 100 μl of Adams-Evans (AE) elution buffer. DNA adequacy was checked by PCR amplification of a 268-bp fragment of the β-globin gene (25) in all the HC2-positive samples that gave negative results at genotyping.
Genotyping assays.
Genotyping was conducted in the molecular laboratories of 4 out of the 9 centers involved in the NTCC trial: Florence (where they also typed the samples from Viterbo), Padua, Trento (where they also typed the samples from Bologna, Imola, and Ravenna), and Turin. Methods were optimized and protocols shared to ensure equivalent accuracy and efficiency of the results.
Genotyping was performed by using the Digene HPV genotyping RH kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions, with slight modifications (26). The assay targets 18 high-risk HPV types (27), which include types assigned by the IARC to group 1 (HPV-16, -18, -31, -33, -35, -39, -45, -51, -52, -56, -58, and -59), group 2A (HPV-68), and group 2B (HPV-26, -53, -66, -73, and -82) (9). Briefly, following a PCR assay with GP5+/GP6+ consensus primers (28), a reverse line blot (RLB) hybridization assay was performed. An HPV-positive control and two negative PCR controls (a purified DNA HPV-negative sample and a DNA-free sample) were included in each PCR run. The sensitivity limit of the assay obtained in our laboratories that participated in the WHO HPV LabNet Proficiency study (29) was 50 copies/5 μl for HPV-16, -18, -31, -33, -35, -45, -56, and -66, and 500 copies/5 μl for HPV-39, -51, -52, -58, -59, -68a, and -68 (ME 180). Ten microliters of PCR biotinylated products was denatured and hybridized at 50°C with type-specific oligonucleotide probes immobilized as parallel lines on nitrocellulose membrane strips. The hybrids were detected with alkaline phosphatase-streptavidin conjugate and substrate (5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium), resulting in a purple precipitate at the positive probe lines. After 30 min, this reaction was stopped by washing the mixture in distilled water. After drying, the strips were analyzed visually from an interpretation grid supplied with the kit; the presence of a clearly visible line was considered to be a positive result.
All samples that remained untyped by this method but which had shown a visible amplicon using gel electrophoresis following PCR with the consensus primers were submitted to further analytical procedures to increase the efficiency of typing, with the aim of identifying the true proportion of false positives for any HPV type. Restriction fragment length polymorphism (RFLP) (30) and direct sequencing procedures were performed for untyped samples, and nested PCR was performed for the samples that remained HPV negative (26) in order to identify the samples with no HPV present.
CIN assessment.
All HC2-positive women were referred to colposcopy except women aged 25 to 34 years at recruitment during phase 1. If HC2 was positive but cytology was normal, the latter group was referred to colposcopy only if HC2 was confirmed as positive or cytology became abnormal after 1 year. All colposcopically suspected areas were biopsied. Biopsy samples of women with a CIN diagnosis were reviewed blindly to test results (31).
Statistical methods.
Each sample was assigned, considering the HPV genotypes that can be detected by any method, to one of five classes: class a if at least one of the types targeted by HC2 was detected, also when in coinfection with other HPV types, class b if no HC2-targeted type was found but at least one of the HPV types classified by the IARC in group 2B was detected, i.e., one that is possibly carcinogenic to humans (HPV-26, -53, -66, -67, -70, -73, and -82) and types with phylogenetic similarity to types with sufficient or limited evidence for cervical cancer in humans (HPV-30, -34, -69, -85, and -97) (9), and when in coinfection with other HPV types, class c if only HPV types not targeted by HC2 and not classified as 2B were present, class d if there was untyped HPV only, and class e if no HPV DNA was detected. The proportion of samples in class a among all HC2-positive samples represented the analytical positive predictive value (PPV) of HC2. Its confidence intervals were computed based on the exact binomial distribution. An approximate analytical specificity was computed with the assumption that all HC2-negative samples were true negatives (32). The number of false positives was computed applying the observed PPV in genotyped samples to all HC2-positive women. Given the high analytical sensitivity of HC2 and the relatively low frequency of HC2-targeted HPV infections in the studied population (4), such an approximation is plausibly very good.
The presence of HC2-targeted types (class a) and of any HPV type (classes a to d) among HC2-positive samples was cross-tabulated with age, RLU/PC ratio, and type of transport medium. In addition, in order to investigate the determinants of cross-reactivity with non-HC2-targeted HPV types, we studied the presence of the HPV types targeted by HC2 (class a) among the samples in which any HPV was detected (classes a to d). Odds ratios adjusted for age, RLU/PC ratio, type of transport medium, and laboratory that performed genotyping were obtained by unconditional logistic regression analysis. In these models, age and the log of the RLU/PC ratio were considered to be linear variables. Wald-type confidence intervals were computed.
The proportion of histologically confirmed CIN2 and CIN3+ detected in each class defined above and the clinical PPV of each class for such lesions were computed. Confidence intervals were computed based on the exact binomial distribution. In addition, in order to understand how much the effect of the type of transport medium and viral load on clinical PPV was the result of analytical inaccuracies, we computed their PPV, both including and excluding analytical false positives.
RESULTS
The overall HC2 positivity rate in the study population was 8.2% (3,033/37,077); the proportion was higher in women aged <35 years (13.57%) (P < 0.0001) and during the first phase of the NTCC trial when PreservCyt transport medium was used (8.8%) (P < 0.0001). Genotyping results are reported in Table 1. Out of 2,920 available HC2-positive DNA samples, 2,310 (79.1%) were positive for one or more of the 13 HPV types targeted by HC2 (class a), while 301 (10.31%) were negative for HC2-targeted types but contained at least one level 2B HPV type (class b), 62 (2.12%) were positive for other HPV types only (class c), 33 (1.13%) remained untyped with all the methods employed (class d), and 214 (7.33%) were negative to all the HPV detection methods applied (class e). Therefore, 610/2,090 (20.9%) samples did not contain HC2-targeted types, although 64.9% of them (396) did contain HPV DNA. The analytical positive predictive value (PPV) of HC2 was 79.1% (95% CI, 77.6 to 80.6). When we also considered the samples with group 2B types and those that were untyped to be true positives, the PPV increased to 90.6% (95% CI, 89.5 to 91.6). The approximate analytical specificity of HC2 for the targeted HPV types was 98.2%. The RLB assay, performed directly following PCR, allowed for the genotyping of 2,383 samples (89.2% of typed samples were in classes a to c), including 2,139/2,310 (94.9%) of those containing HC2-targeted types. Performing nested PCR and/or the other molecular procedures (see Materials and Methods) allowed for the genotyping of 290 additional samples.
Table 1.
Distribution of the HC2-positive samples according to the genotypes detected overall and by typing method
| Genotype class and sample description | No. (% of total) of samples in each class | No. (% of total) detected in each class with: |
|
|---|---|---|---|
| Digene HPV genotyping RH kita | Supplemental typing methodsb | ||
| a: Samples with HC2-targeted HPV types | 2,310 (79.11) | 2,193 (94.9) | 117 (5.1) |
| b: Samples with types in the IARC group 2B only | 301 (10.31) | 187 (62.1) | 114 (37.9) |
| c: Samples with other HPV types only | 62 (2.12) | 3 (4.8) | 59 (95.2) |
| d: Untyped samples | 33 (1.13) | 0 (0) | 33 (1.13) |
| e: Samples with no HPV DNA detected | 214 (7.33) | 11 (5.1) | 203 (94.9) |
| Total no. of samples | 2,920 | 2,394 (82.0) | 526 (18.0) |
RLB, reverse line blot hybridization following PCR with GP5+/GP6+ consensus primers.
Supplemental typing methods include RLB-negative samples retested for typing detection, RLB following nested PCR, restriction fragment length polymorphism, or direct sequencing. See Materials and Methods for details.
Among the 363 samples with only non-HC2-targeted identified HPV types (classes b and c), 379 infections from 24 different HPV types were detected. HPV-53 (20.05%), HPV-66 (35.09%), and HPV-70 (11.08%) were by far the most frequently detected and altogether represented 66.2% of such infections, while none of the other types (HPV-6, -32, -40, -42, -43, -44, -54, -61, -62, -67, -69, -73, -74, -81, -82, -83, -84, -87, -89, -90, and -91) accounted for >7% of them. In addition, 68.3% of the samples with only non-HC2-targeted HPV types identified contained HPV-53, HPV-66, or HPV-70, either alone or in coinfection with other non-HC2-targeted types.
The proportion of samples with HC2-targeted types, and with any HPV type among HC2-positive samples, significantly decreased with increasing age at recruitment and strongly with the use of the PreservCyt over the STM transport solution (Tables 2 and 3). These proportions also significantly and strongly increased with increasing viral load (RLU/PC). They were very low, with RLU/PC values close to the cutoff (RLU/PC <2, but still for values <4), particularly when the PreservCyt solution was used. Using the PreservCyt solution, <60% of the samples with an RLU/PC ratio of <4 contained HC2-targeted HPV types (Table 2), and a higher proportion of samples with no HPV detected (class e) was found than when the STM solution was used (Table 3). The ORs for all determinants were less extreme when the presence of HC2-targeted types was the endpoint and more extreme when the presence of any HPV DNA was the endpoint. However, this change was particularly large for the type of transport medium: the OR was 1.52 (95% CI, 1.25 to 1.84) for the presence of HC2-targeted types and 2.93 (95% CI, 2.04 to 3.91) for the presence of any HPV DNA (Table 3). Conversely, the proportion of samples with HC2-targeted types among the samples with any HPV type were only affected by viral load, while it was not associated with age and type of transport medium (Tables 2 and 3). Results related to type of transport medium were virtually unchanged when excluding the centers that collected samples only in phase 2 (data not shown).
Table 2.
Variability in the detection of HPV types according to age, type of collection medium, and sample viral load
| Study variables | Total no. of samples | No. (%) of HC2-targeted types present | No. (%) of any HPV type present | No. of HC2-targeted types present among samples containing any HPV type (%) |
|
|---|---|---|---|---|---|
| No. of samples tested | No. present (% of total tested) | ||||
| Age (yr) | |||||
| 25–34 | 1,310 | 1,092 (83.4) | 1,256 (95.9) | 1,256 | 1,092 (86.9) |
| 35–44 | 921 | 721 (78.3) | 852 (95.5) | 852 | 721 (84.6) |
| 45–60 | 689 | 497 (72.1) | 598 (86.8) | 598 | 497 (83.1) |
| Collection medium type and characteristics | |||||
| PreservCyt (all samples) | 1,264 | 945 (74.7) | 1,126 (89.1) | 1,126 | 945 (83.9) |
| STM (all samples) | 1,656 | 1,365 (82.4) | 1,580 (95.4) | 1,580 | 1,365 (86.4) |
| PreservCyt | |||||
| RLU/PC ratioa | |||||
| <2 | 227 | 126 (55.5) | 167 (73.6) | 167 | 126 (75.5) |
| 2–3.99 | 141 | 83 (58.9) | 99 (70.2) | 99 | 83 (83.8) |
| 4–9.99 | 155 | 106 (68.4) | 134 (86.5) | 134 | 106 (79.1) |
| ≥10 | 741 | 630 (85.0) | 726 (97.9) | 726 | 630 (86.8) |
| STM | |||||
| RLU/PC ratio | |||||
| <2 | 275 | 197 (71.6) | 237 (88.2) | 237 | 197 (83.12) |
| 2–3.99 | 191 | 141 (73.8) | 173 (90.6) | 173 | 141 (81.5) |
| 4–9.99 | 174 | 135 (77.6) | 167 (96.0) | 167 | 135 (80.8) |
| ≥10 | 1,016 | 892 (87.8) | 1,003 (98.7) | 1,003 | 892 (88.9) |
RLU/PC ratio is the ratio between the specimen relative light units (RLU) and the positive cutoff (PC) value of 1 pg/ml of HPV DNA, a relative measure of the HPV viral load in the sample.
Table 3.
Association of HPV types detected with age, type of transport medium, and sample viral load
| HPV type(s) present and group characteristics | Odds ratio (95% CI)a | P |
|---|---|---|
| HC2 targeted types present | ||
| Age at recruitmentb | 0.99 (0.98–1.00) | 0.0296 |
| Viral loadc | 1.35 (1.30–1.42) | <0.0001 |
| Transport mediumd | 1.52 (1.25–1.84) | <0.0001 |
| Any HPV type present | ||
| Age at recruitment | 0.97 (0.96–0.99) | 0.0002 |
| Viral load | 2.01 (1.70–2.26) | <0.0001 |
| Transport medium | 2.83 (2.04–3.91) | <0.0001 |
| HC2-targeted types present among samples containing any HPV type | ||
| Age at recruitment | 1 (0.99–1.01) | 0.8198 |
| Viral load | 1.21 (1.15–1.27) | <0.0001 |
| Transport medium | 1.15 (0.92–1.44) | 0.21 |
OR was adjusted reciprocally and for a laboratory that performed genotyping by unconditional logistic regression.
Age at recruitment represents a 1-year increase.
Viral load represents 1-log RLU/PC increase. RLU/PC ratio is the ratio between the specimen relative light units (RLU) and the positive cutoff (PC) of 1 pg/ml of HPV DNA, a relative measure of the HPV viral load in the sample.
Transport medium is STM versus PreservCyt.
Histologically confirmed high-grade lesions were detected at recruitment in 204 of the HC2-positive studied women, 117 with CIN2 and 87 with CIN3 (Table 4). Ten of these occurred in women without HC2-targeted types. Of these 10 women (including 2 with CIN3), four had HPV types classified by IARC as group 2B, another four had only other HPV types, and two had no HPV DNA detected. If there were no analytical false-positive results, 194 high-grade lesions would have been detected, 109 being CIN2 and 85 being CIN3; therefore, 4.9% (95% CI, 2.38% to 8.88%) of CIN2+ and 2.3% of CIN3 (95% CI, 0.2% to 8.1%) would have been missed. In this case, however, the clinical PPV for histologically confirmed CIN2+ of HC2 positivity would have increased from 7.0% (95% CI, 6.1 to 8.0) to 8.4% (95% CI, 7.3 to 9.6). The PPVs for CIN2+ (1.6% [95% CI, 0.8 to 3.0]) and especially for CIN3 (0.3% [95% CI, 0.004 to 1.2]) were very low in false-positive cases and surprisingly were lower among women with group 2B types than in women with only noncarcinogenic types (Table 4). Two women with CIN2 and none with CIN3 had samples containing HPV-53 out of 76 women with infections by that type. Almost all the differences in PPVs for CIN2+ between women with samples taken in PreservCyt versus STM disappeared when excluding the cases without HC2-targeted types. However, this exclusion affected only marginally the difference in PPVs between women with an RLU/PC ratio of <4 versus a ratio of ≥4 (Tables 4 and 5).
Table 4.
High-grade cervical lesions and PPV in HC2-positive women according to the results of genotyping
| Genotype class and sample definition | No. of samples | No. of samples with CIN grade: |
PPV by CIN type (% [95% CI])a |
||
|---|---|---|---|---|---|
| 2 | 3 | CIN2+ | CIN3 | ||
| a: Samples with HC2-targeted HPV types | 2,310 | 109 | 85 | 8.4 (7.3–9.6) | 3.6 (3.0–4.5) |
| b: Samples with just types in the IARC group 2B | 301 | 3 | 1 | 1.3 | 0.3 |
| c: Samples with just other HPV types | 62 | 3 | 1 | 6.5 | 1.6 |
| d: Untyped samples | 33 | 0 | 0 | 0.0 | 0.0 |
| e: Samples with no HPV detected | 214 | 2 | 0 | 0.9 | 0.0 |
| Total samples without HC2-targeted types | 610 | 8 | 2 | 1.6 (0.8–3.0) | 0.3 (0.004–1.2) |
| Total HC2-positive samples | 2,920 | 117 | 87 | 7.0 (6.1–8.0) | 2.8 (2.4–3.7) |
95% confidence intervals (CIs) are based on the exact binomial distribution.
Table 5.
PPV for CIN2+ samples by type of transport medium and RLU/PC ratio, when considering all HC2-positive women or only those with analytical true positives
| Target group | PreservCyt samples |
STM samples |
RLU/PC ratio<4 |
RLU/PC ratio ≥4 |
||||
|---|---|---|---|---|---|---|---|---|
| No. of CIN2+/total samples | PPV (% [95% CI]) | No. of CIN2+/total samples | PPV (% [95% CI]) | No. of CIN2+/total samples | PPV (% [95% CI]) | No. of CIN2+/total samples | PPV (% [95% CI]) | |
| Women with HC2-positive samples | 80/1,264 | 6.3 (5.1–7.8) | 124/1,656 | 7.5 (6.3–8.9) | 13/834 | 1.56 (0.8–2.7) | 191/2,086 | 9.2 (8.0–10.5) |
| Women with HC2-targeted types | 76/945 | 8.0 (6.4–10.0) | 118/1,365 | 8.6 (7.2–10.3) | 9/547 | 1.65 (0.8–3.1) | 185/1,763 | 10.5 (9.1–12.0) |
DISCUSSION
In our study, 20.9% of 2,920 HC2-positive samples did not contain HC2-targeted HPV types and can be considered to be analytical false positives. This happened despite an approximately 98.2% analytical specificity because of the low prevalence of targeted types in the study population. About half of those samples contained HPV types with limited (IARC group 2B) evidence of carcinogenicity. However, the clinical value of these positive tests is small, given the low probability of detecting CIN2 and especially CIN3 in these women, including those with HPV-53. Overall, the loss in clinical sensitivity that resulted by missing the HPV types not targeted by HC2 would be small, in line with previous reports (15) suggesting that efforts to increase the number of targeted HPV types (i.e., HPV-53, HPV-66, and others) in HPV screening tests would not significantly improve clinical sensitivity. In addition, the probability of progression of these CINs is unknown, and histological false positives cannot be excluded (31, 33). Conversely, improvements in analytical specificity would have a nonnegligible impact in reducing unneeded triage and/or colposcopy.
Of the 610 false-positive samples that contained non-HC2-targeted HPV types, 64.9% contained HPV DNA and could therefore possibly result from cross hybridization, while the 30.1% of samples in which no HPV DNA could be detected even following intensive search were plausibly caused by other technical problems. However, the proportion of false positives that is actually caused by cross-hybridization might be slightly <64.9% because false positivity caused by non-cross-hybridization problems might also have occurred in samples containing non-HC2-targeted HPV types.
Also, in our sample series, the large majority of cross-hybridizations occurred with types HPV-53, HPV-66, and HPV-70, which are phylogenetically related to the HPV types included in HC2 (34, 35), which is consistent with observations in previous studies (10–13, 16, 20).
Increasing age (as was also observed in the ARTISTIC study [20]), the use of the PreservCyt transport medium, and a low RLU/PC ratio were significantly associated with false-positive HC2 results. The two latter trends especially have practical relevance.
The use of the PreservCyt transport medium instead of STM does not seem to be a determinant of cross-hybridization but is plausibly an important cause of the remaining false-positive results obtained with HC2. Indeed, in this study, PreservCyt use was not associated with a lack of HC2-targeted types among samples that contained some HPV DNA, while it showed a very strong association with the absence of any HPV DNA. In the ARTISTIC study where PreservCyt was used, no HC2-targeted type was detected in some 30% of samples from HC2-positive women (20) and, interestingly, in most of them, no HPV could be detected (with a less-intensive approach than ours). PreservCyt samples need preanalytical treatment to achieve suitability for performance with the HC2 test. Such a preanalytical step, when carried out using a manual conversion system (sample conversion kit; Digene-Qiagen), can be critical since residual traces of undenatured DNA may generate results in a gray zone with RLU/PC ratios between 1 and 2.5 and an increase in borderline false-positive results (19, 23).
New technologies that allow raw DNA extraction (QIAsymphony, Qiagen) are now available to escape the manual conversion step that precedes the HC2 assay. PreservCyt allows for the performance of HPV testing and liquid-based cytology on the same material, which represents an advantage if cytological triage for HPV-positive women or cotesting is applied. Its effects on false-positive HC2 results should, however, be considered when the manual conversion procedure is used. Among women aged >35 years, a lower HC2 prevalence was found in phase 2 of NTCC study when STM was used (5.8%) than in phase 1 when PreservCyt was used (7.1%), while the PPV of HC2 positivity for high-grade CIN was higher in phase 2 (7.2%) than in phase 1 (6.6%) (2, 4). These differences might largely be the result of a lower proportion of analytical false-positive results with the use of STM. Indeed, in our study, the differences in PPV by the type of transport medium almost disappeared when excluding women with false-positive analytical results.
A decreasing RLU/PC ratio was strongly associated with increasing frequency of false-positive results. As a low RLU/PC ratio was associated with both lack of any HPV and the absence of HC2-targeted types among samples with HPV, both cross-hybridization and other mechanisms seem to be relevant. In all RCTs that used HC2, increasing the cutoff from 1 to 2 or 4 RLU/PC decreased the sensitivity only marginally, while the proportion of HC2-positive women without high-grade CIN was strongly reduced. Part of this effect might be due to the fact that about 40% of cases collected in PreservCyt and 30% of those collected in STM with an RLU/PC ratio between 1 and 4 did not contain the test-targeted HPV types. However, most of this effect seems to be related to viral load. Indeed, in our study, the differences in PPV for high-grade CIN between women with high and low RLU/PC ratios remained almost unchanged when excluding analytical false positives. The HC2 manufacturer currently recommends a retesting procedure of PreservCyt samples when HPV-positive results fall into a gray zone with RLU/PC ratios between 1 and 2.5, with an indication to repeat the HC2 test once if the second result is concordant with the first, or twice if discordant. In our setting, the gray zone was not considered since the NTCC trial started in 2002, when the retesting procedure was not yet recommended. This procedure has also an impact on costs, which is worthy of consideration.
This study has strengths and limitations. The studied samples are highly representative of those coming from a general screening population. In addition, intensive analytical efforts with the hierarchical application of different approaches was done by laboratories conducting internal-external quality assurance, in order to identify all infections present. However, we cannot exclude the possibility that some of the disagreements between the HC2 and genotyping were not due to HC2 false positives but to analytical inaccuracies of the genotyping process (problems in DNA preservation are not plausible, as all negative samples were positive for beta-globin amplification). The potential underestimation of multiple infections also has to be taken into account due to the use of GP5+/GP6+ primers in the assay we employed (36). These factors might have led to an overestimation of HC2 false positives and possibly to misclassifications in their subcategories. Nevertheless, as data showed a strong association of false positives and subgroups with plausible causal determinants, it is unlikely that the above-mentioned overestimation and misclassifications would have a significant impact on results.
Confounding by plausible determinants was controlled by the adjustment of ORs through logistic regression analysis. Some centers were considered only in phase 2 (when STM was used); however, results on type of transport medium were virtually unchanged when excluding these centers.
Finally, it must be kept in mind that only the impact of HC2 cross-reactivity on the PPV for the detection of prevalent CIN2+, but not its prognostic value on subsequent lesions, was considered.
In conclusion, in the NTCC study, some 20% of HC2-positive samples did not contain the targeted HPV types. About two-thirds of them resulted from cross-hybridization, especially with HPV-53, HPV-66, and HPV-70. New validated screening tests include in their panel HPV-66, and positivity found with its presence probably compensates in comparative studies for the cross-hybridization of HC2. However, including the detection of HPV-53, HPV-66, and HPV-70 results in very little gains in sensitivity for high-grade CIN. Using STM instead of PreservCyt (followed by a manual conversion procedure) as the transport medium and/or increasing the cutoff as previously suggested (37) reduces the analytical false-positive rate and increases the PPV for high-grade CIN.
ACKNOWLEDGMENTS
The study was supported by the Italian Ministry of Health (special project “Valutazione nuove tecnologie per lo screening del cervicocarcinoma–follow up”) and the European Union (PREHDICT project FP5 grant agreement no. 242061).
The Italian NTCC Working Group collaborators are: in Turin, N. Segnan, R. Rizzolo, and P. Giubilato (CPO Piemonte), B. Ghiringhello (Unit of Pathology, OIRM S. Anna), A. Sapino and M. G. Accinelli (Unit of Pathology, University of Turin), and G. Maina and L. Pasero (Centre for Cancer Early Diagnosis and Treatment, OIRM S. Anna); in Trento, P. Dalla Palma and E. Polla (Unit of Pathology, Ospedale di Trento); in Veneto, L. Baboci, R. Trevisan, M. Zorzi, C. Fedato, and S. Baracco (Istituto Oncologico Veneto–IRCCS, Padova), E. Insacco, D. Minucci, and M. Matteucci (Unity of Gynaecology, Azienda Ospedaliera di Padova), G. L. Onnis (Department of Pathology, University of Padua), and R. Colombari (Unit of Pathology, Ospedale di S. Bonifacio, AUSL Verona); in Emilia-Romagna, M. Manfredi (Centro Screening, AUSL Bologna), G. P. Casadei (Unit of Pathology, Ospedale Maggiore, AUSL Bologna), G. Collina (Unit of Pathology Ospedale Bellaria, AUSL Bologna), P. Schincaglia, M. Serafini, and B. Vitali (Centro Prevenzione Oncologica, AUSL Ravenna), M. Aldi (Unit of Pathology, Presidio Ospedaliero di Faenza, AUSL Ravenna), S. Folicaldi, R. Nannini, G. Galanti, and M. De Lillo (Unit of Pathology, Presidio Ospedaliero, AUSL di Imola), and C. Naldoni (Centro di Riferimento screening, Assessorato alla Sanità, Regione Emilia-Romagna, Bologna); in Florence, M. Confortini, A. Iossa, C. Sani, S. Ciatto, M. P. Cariaggi, S. Cecchini, M Zappa (ISPO Florence), and G. L. Taddei (Unit of Pathology, University of Florence); and in Lazio, S. Brezzi, P. Raggi, and E. Gomes (Local Health Unit, Viterbo, Italy), A. Pellegrini and M. L. Schiboni (Ospedale S. Giovanni, Rome), and P. Borgia and F. Chini (ASP Lazio).
The authors disclose no conflicts of interest.
Footnotes
Published ahead of print 26 June 2013
REFERENCES
- 1.Cuzick J, Mayrand MH, Ronco G, Snijders P, Wardle J. 2006. Chapter 10: new dimensions in cervical cancer screening. Vaccine 24(Suppl 3): S3/90–S3/97 [DOI] [PubMed] [Google Scholar]
- 2.Ronco G, Segnan N, Giorgi-Rossi P, Zappa M, Casadei G, Crozzi F, Dalla Palma P, Del Mistro A, Focalidi S, Gillio-Tos A, Nardo G, Naldoni C, Schincaglia P, Zorzi M, Confortini M, Cuzick J, the Working Group NTCC 2006. Human papillomavirus testing and liquid-based cytology in primary cervical screening: results at recruitment from NTCC randomized controlled trial. J. Natl. Cancer Inst. 98:765–774. 10.1093/jnci/djj209 [DOI] [PubMed] [Google Scholar]
- 3.Mayrand MH, Duarte-Franco E, Rodrigues I, Walter SD, Hanley J, Ferenczy A, Ratnam S, Coutlée F, Franco EL; Canadian Cervical Cancer Screening Trial Study Group 2007. Human papillomavirus DNA versus Papanicolaou screening tests for cervical cancer. N. Engl. J. Med. 357:1579–1588 [DOI] [PubMed] [Google Scholar]
- 4.Ronco G, Giorgi-Rossi P, Carozzi F, Confortini M, Dalla Palma P, Del Mistro A, Gillio-Tos A, Minucci D, Naldoni C, Rizzolo R, Schincaglia P, Volante R, Zappa M, Zorzi M, Cuzick J, Segnan N; New Technologies for Cervical Cancer Screening Working Group 2008. Results at recruitment from a randomized controlled trial comparing human papillomavirus testing alone with conventional cytology as the primary cervical cancer screening test. J. Natl. Cancer Inst. 100:492–501 [DOI] [PubMed] [Google Scholar]
- 5.Leinonen M, Nieminen P, Kotaniemi-Talonen L, Malila N, Tarkkanen J, Laurila P, Anttila A. 2009. Age-specific evaluation of primary human papillomavirus screening vs conventional cytology in a randomized setting. J. Natl. Cancer Inst. 101:1612–1623 [DOI] [PubMed] [Google Scholar]
- 6.Kitchener HC, Almonte M, Thomson C, Wheeler P, Sargent A, Stoykova B, Gilham C, Baysson H, Roberts C, Dowie R, Desai M, Mather J, Bailey A, Turner A, Moss S, Peto J. 2009. HPV testing in combination with liquid-based cytology in primary cervical screening (ARTISTIC): a randomised controlled trial. Lancet Oncol. 10:672–682 [DOI] [PubMed] [Google Scholar]
- 7.Meijer CJ, Berkhof J, Castle PE, Hesselink AT, Franco EL, Ronco G, Arbyn M, Bosch FX, Cuzick J, Dillner J, Heideman DA, Snijders PJ. 2009. Guidelines for human papillomavirus DNA test requirements for primary cervical cancer screening in women 30 years and older. Int. J. Cancer 124:516–520 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Giorgi-Rossi P, Franceschi S, Ronco G. 2012. HPV prevalence and accuracy of HPV testing to detect high-grade cervical intraepithelial neoplasia. Int. J. Cancer 130:1387–1394 [DOI] [PubMed] [Google Scholar]
- 9.Bouvard V, Baan R, Straif K, El Ghissassi F, Benbrahim-Tallaa L, Guha N, Freeman C, Galichet L, Gogliano V, on behalf of the WHO IARC in Cancer Monograph Working Group 2009. A review of human carcinogens. Part B: biological agents. Lancet Oncol. 10:321–322 [DOI] [PubMed] [Google Scholar]
- 10.Lindemann ML, Dominguez MJ, de Antonio JC, Sandri MT, Tricca A, Sideri M, Khiri H, Ravet S, Boyle S, Aldrich C, Halfon P. 2012. Analytical comparison of the cobas HPV test with Hybrid Capture 2 for the detection of high-risk HPV genotypes. J. Mol. Diagn. 14:65–70 [DOI] [PubMed] [Google Scholar]
- 11.Castle PE, Solomon D, Wheeler CM, Gravitt PE, Wacholder S, Shiffman M. 2008. Human papillomavirus genotype specificity of Hybrid Capture 2. J. Clin. Microbiol. 46:2595–2604 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Safaeian M, Herrero R, Hildesheim A, Quint W, Freer E, Van Doorn LJ, Porras C, Silva S, González P, Bratti MC, Rodriguez AC, Castle P, Costa Rican Vaccine Trial Group 2007. Comparison of the SPF10-LiPA system to the Hybrid Capture 2 Assay for detection of carcinogenic human papillomavirus genotypes among 5,683 young women in Guanacaste, Costa Rica. J. Clin. Microbiol. 45:1447–1454 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Schiffman M, Wheeler CM, Dasgupta A, Solomon D, Castle PE, ALTS Group 2005. A comparison of a prototype PCR assay and hybrid capture 2 for detection of carcinogenic human papillomavirus DNA in women with equivocal or mildly abnormal papanicolaou smears. Am. J. Clin. Pathol. 124:722–732 [DOI] [PubMed] [Google Scholar]
- 14.Poljak M, Marin IJ, Seme K, Vince A. 2002. Hybrid Capture II HPV test detects at least 15 human papillomavirus genotypes not included in its current high-risk probe cocktail. J. Clin. Virol. 25(Suppl 3):S89–S97. 10.1016/S1386-6532(02)00187-7 [DOI] [PubMed] [Google Scholar]
- 15.Castle PE, Schiffman M, Burk RD, Wacholder S, Hildesheim A, Herrero R, Bratti MC, Sherman ME, Lorincz A. 2002. Restricted cross-reactivity of hybrid capture 2 with nononcogenic human papillomavirus types. Cancer Epidemiol. Biomarkers Prev. 11:1394–1399 [PubMed] [Google Scholar]
- 16.Schneede P, Hillemanns P, Ziller F, Hofstetter A, Stockfleth E, Arndt R, Meyer T. 2001. Evaluation of HPV testing by Hybrid Capture II for routine gynecologic screening. Acta Obstet. Gynecol. Scand. 80:750–752 [DOI] [PubMed] [Google Scholar]
- 17.Terry G, Ho L, Londesborough P, Cuzick J, Mielzynska-Lohnas I, Lorincz A. 2001. Detection of high-risk HPV types by the hybrid capture 2 test. J. Med. Virol. 65:155–162 [PubMed] [Google Scholar]
- 18.Yamazaki H, Sasagawa T, Basha W, Segawa T, Inoue M. 2001. Hybrid capture-II and LCR-E7 PCR assays for HPV typing in cervical cytologic samples. Int. J. Cancer 94:222–227 [DOI] [PubMed] [Google Scholar]
- 19.Peyton CL, Schiffman M, Lörincz AT, Hunt WC, Mielzynska I, Bratti C, Eaton S, Hildesheim A, Morera LA, Rodriguez AC, Herrero R, Sherman ME, Wheeler CM. 1998. Comparison of PCR- and hybrid capture-based human papillomavirus detection systems using multiple cervical specimen collection strategies. J. Clin. Microbiol. 36:3248–3254 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sargent A, Bailey A, Almonte M, Turner A, Thomson C, Peto J, Desai M, Mather J, Moss S, Roberts C, Kitchener HC, ARTISTIC Study Group 2008. Prevalence of type-specific HPV infection by age and grade of cervical cytology: data from the ARTISTIC trial. Br. J. Cancer 98:1704–1709 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Dalstein V, Merlin S, Bali C, Saunier M, Dachez R, Ronsin C. 2008. Analytical evaluation of the PapilloCheck test, a new commercial DNA chip for detection and genotyping of human papillomavirus. J. Virol. Methods 156:77–83 [DOI] [PubMed] [Google Scholar]
- 22.Seme K, Fujs K, Kocjan BJ, Poljak M. 2006. Resolving repeatedly borderline results of Hybrid Capture 2 HPV DNA test using polymerase chain reaction and genotyping. J. Virol. Methods 134:252–256 [DOI] [PubMed] [Google Scholar]
- 23.Carozzi FM, Del Mistro A, Confortini M, Sani C, Puliti D, Trevisan R, De Marco L, Gillio-Tos A, Girlando S, Dalla Palma P, Pellegrini A, Schiboni ML, Crucitti P, Pierotti P, Vignato A, Ronco G. 2005. Reproducibility of HPV DNA testing by Hybrid Capture 2 in a screening setting. Am. J. Clin. Pathol. 124:716–721 [DOI] [PubMed] [Google Scholar]
- 24.Ronco G, Giorgi-Rossi P, Carozzi F, Confortini M, Dalla Palma P, Del Mistro A, Ghiringhello B, Girlando S, Gillio-Tos A, De Marco L, Naldoni C, Pierotti P, Rizzolo R, Schincaglia P, Zorzi M, Zappa M, Segnan N, Cuzick J, New Technologies for Cervical Cancer screening (NTCC) Working Group 2010. Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: a randomised controlled trial. Lancet Oncol. 11:249–257 [DOI] [PubMed] [Google Scholar]
- 25.Dong SM, Pai SI, Rha SH, Hildesheim A, Kurman RJ, Schwartz PE, Mortel R, McGowan L, Greenberg MD, Barnes WA, Sidransky D. 2002. Detection and quantitation of human papillomavirus DNA in the plasma of patients with cervical carcinoma. Cancer Epidemiol. Biomarkers. Prev. 11:3–6 [PubMed] [Google Scholar]
- 26.Carozzi F, Tornesello ML, Burroni E, Loquercio G, Carillo G, Angeloni C, Scalisi A, Macis R, Chini F, Buonaguro FM, Giorgi Rossi P, HPV Prevalence Italian Working Group 2010. Prevalence of human papillomavirus types in high-grade cervical intraepithelial neoplasia and cancer in Italy. Cancer Epidemiol. Biomarkers Prev. 19:2389–2400 [DOI] [PubMed] [Google Scholar]
- 27.Geraets DT, Heideman DA, de Koning MN, Snijders PJ, Meijer CJ, van Doorn LJ, Quint WG. 2009. High genotyping concordance between the Digene HPV genotyping RH test and the reverse line blot genotyping assay on GP5+/6+-PCR products. J. Clin. Virol. 46(Suppl 3):S16–S20. 10.1016/S1386-6532(09)70296-3 [DOI] [PubMed] [Google Scholar]
- 28.van den Brule AJ, Pol R, Fransen-Daalmeijer N, Schouls LM, Meijer CJ, Snijders PJ. 2002. GP5+/6+ PCR followed by reverse line blot analysis enables rapid and high-throughput identification of human papillomavirus genotypes. J. Clin. Microbiol. 40:779–787 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Eklund C, Forslund O, Wallin KL, Zhou T, Dillner J, WHO Human Papillomavirus Laboratory Network 2012. The 2010 global proficiency study of human papillomavirus genotyping in vaccinology. J. Clin. Microbiol. 50:2289–2298 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Nobre RJ, de Almeida LP, Martins TC. 2008. Complete genotyping of mucosal human papillomavirus using a restriction fragment length polymorphism analysis and an original typing algorithm. J. Clin. Virol. 42:13–21 [DOI] [PubMed] [Google Scholar]
- 31.Dalla Palma P, Giorgi-Rossi P, Collina G, Buccoliero AM, Ghiringhello B, Lestani M, Onnis GL, Aldovini D, Galanti G, Casadei GP, Aldi M, Gomes V, Giubilato V, Ronco G, NTCC Pathology Group 2008. The risk of false-positive histology according to the reason for colposcopy referral in cervical cancer screening: a blind revision of all histological lesions found in the NTCC trial. Am. J. Clin. Pathol. 129:75–80 [DOI] [PubMed] [Google Scholar]
- 32.Arbyn M, Ronco G, Cuzick J, Wentzensen N, Castle PE. 2009. How to evaluate emerging technologies in cervical screening? Int. J. Cancer 125:2489–2496 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Castle PE, Cox JT, Jeronimo J, Solomon D, Wheeler CM, Gravitt PE, Schiffman M. 2008. An analysis of high-risk human papillomavirus DNA-negative cervical precancers in the ASCUS-LSIL Triage Study (ALTS). Obstet. Gynecol. 111:847–856 [DOI] [PubMed] [Google Scholar]
- 34.Schiffman M, Herrero R, Desalle R, Hildesheim A, Wacholder S, Rodriguez AC, Bratti MC, Sherman ME, Morales J, Guillen D, Alfaro M, Hutchinson M, Wright TC, Solomon D, Chen Z, Schussler J, Castle PE, Burk RD. 2005. The carcinogenicity of human papillomavirus types reflects viral evolution. Virology 337:76–84 [DOI] [PubMed] [Google Scholar]
- 35.Menzo S, Ciavattini A, Bagnarelli P, Marinelli K, Sisti S, Clementi M. 2008. Molecular epidemiology and pathogenic potential of underdiagnosed human papillomavirus types. BMC Microbiol. 8:112. 10.1186/1471-2180-8-112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Schmitt M, Dondog B, Waterboer T, Pawlita M, Tommasino M, Gheit T. 2010. Abundance of multiple high-risk human papillomavirus (HPV) infections found in cervical cells analyzed by use of an ultrasensitive HPV genotyping assay. J. Clin. Microbiol. 48:143–149 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Rebolj M, Bonde J, Njor SH, Lynge E. 2011. Human papillomavirus testing in primary cervical screening and the cut-off level for hybrid capture 2 tests: systematic review. BMJ 342:d2757. 10.1136/bmj.d2757 [DOI] [PMC free article] [PubMed] [Google Scholar]