HPV genotype-specific distribution and attributable risk in cervical intraepithelial neoplasia in a referral population with a history of LSIL

Abstract

BACKGROUND AND OBJECTIVE:

CINtec PLUS and cobas HPV tests (Roche) were previously ascertained for triaging an LSIL referral population [1]. As part of this study, genotype-specific distribution and attributable risk of high-risk (HR)-HPV in cervical intraepithelial neoplasia (CIN) were determined.

METHODS:

Archived cervical specimens in ThinPrep PreservCyt (Hologic Inc) from the LSIL referral population ( $n=$ 533) were genotyped using the Anyplex II HPV HR test (Anyplex, Seegene Inc). Since the study specimens had been in storage in ambient temperature for 31–47 months since collection, Anyplex results were compared with that of the initial cobas testing of fresh specimens to validate the suitability and stability of specimens for the present study.

RESULTS:

Overall, Anyplex test was positive in 63% (336/533) vs. 55.7% (297/533) for cobas test. Anyplex test performed identical to cobas test identifying 93.2% (82/88) of $\geqslant$ CIN2/adenocarcinoma in situ (AIS). Anyplex test detected genotypes 16/18 in 15.7% (36/230) $\leqslant$ CIN1 vs. 45.5% (40/88) $\geqslant$ CIN2/AIS; the corresponding figures were 13.5% (31/230) and 45.5% (40/48) for the cobas test. Genotype 16 showed increasing attribution, 13.2% in CIN1, 27.1% in CIN2 and 40% in CIN3/AIS. Of the 12 other high-risk (OHR) types collectively identified by cobas, Anyplex test specifically detected, in decreasing order, genotypes 51, 31, 35, 56, 39, and 45 as the most frequent types, often in multiple-type infections, in 64.8% $\geqslant$ CIN2. Regardless, estimated attribution was evident for each of the 12 OHR types in $\geqslant$ CIN2. Multiple-type infections were more frequent than single-type infections in all CIN grades.

CONCLUSIONS:

Attributable risk of all HR-HPV genotypes targeted by both Anyplex and cobas tests was evident in $\geqslant$ CIN2/AIS Testing for these genotypes in HPV primary cervical screening and cytology triage could identify those at increased risk of cervical cancer and also be beneficial in the management of LSIL referral populations.

Keywords

HPV genotype-specific distribution in cervical intraepithelial neoplasia (CIN)HPV genotype-specific attributable risk in CIN Anyplex II HPV HR test cobas HPV test ThinPrep PreservCyt stability for HPV DNA testing

1. Introduction

Persistent infection with oncogenic high-risk human papillomaviruses (HR-HPV) leads to transforming infection and subsequent lesion progression, the underlying precursor of cervical cancer [2, 3]. Testing for HR-HPV is more sensitive and efficient than Papanicolaou cytology for detecting pre-cancer cervical lesions and cancer [4, 5, 6, 7], and HR-HPV testing has been recommended as a standalone screening test or in conjunction with cytology in primary cervical cancer screening, and for triaging women having atypical squamous cells of undetermined significance (ASCUS) abnormality in cytology screening [8, 9, 10, 11]. Consequently, HR-HPV testing is increasingly adopted in primary cervical screening and being used for ASCUS triage in jurisdictions where cytology remains the primary screening method. Also, there is evidence that HR-HPV testing could be useful for triaging women having low-grade squamous intraepithelial lesion (LSIL) cytology [1, 12, 13, 14].

Many commercially available HPV tests target 14 genotypes, 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68, which are recognized as HR oncogenic types associated with cervical cancer. Some of these tests also allow for simultaneous partial or full genotyping in a single analysis, especially identifying genotypes 16/18 which account for most of the cervical cancer worldwide [10, 15, 16, 17, 18, 19]. Currently available next generation HR-HPV tests thus aid in the determination of a genotype-specific risk threshold and this has led to the US guidelines recommending direct referral to colposcopy for women testing positive for genotypes 16/18 in HPV primary screening [20]. This approach could also be helpful in ASCUS/LSIL-HPV triage.

We previously carried out a two-year prospective study to determine the application of CINtec PLUS dual immunostaining and the cobas HPV test (Roche Diagnostics) for triaging Canadian women referred to colposcopy with a history of LSIL cytology [LSIL triage study; 1,14] Cervical specimens collected in ThinPrep PreservCyt (Hologic Inc) cytology medium at patient enrollment for the above LSIL triage study were used in the present expanded genotyping study. The cobas HPV test detects the 14 HR types listed above, and being a partial genotyping test, it identified genotypes 16/18 specifically in 28.1% of the above LSIL triage study population overall, and in 46.3% of those having cervical intraepithelial neoplasia grade 2 or worse ( $\geqslant$ CIN2); 12 other high-risk (OHR) types were detected collectively in 48.1% [1]. Although the 12 OHR types are known to be far more prevalent than types 16/18 in ASCUS and LSIL cytology and low-grade CIN (CIN1), in terms of genotype-specific attributable risk in cervical cancer, types 16/18 account for about 70%, and the rest are mostly associated with the 12 OHR types [15, 16, 17, 21]. Knowledge of the distribution of specific HPV types in CIN is important in HPV-based screening and triage and could be helpful in patient management and follow-up strategies as well as in post-treatment assessment. The present study was conducted as part of the above LSIL triage study with an objective to specifically identify individual genotypes and to elucidate their distribution and attributable rate of detection in $\leqslant$ CIN1 vs. $\geqslant$ CIN2 in the LSIL referral population.

The Anyplex II HPV HR assay (Anyplex; Seegene Inc) is a multiplex real-time PCR amplification test and detects DNA of the same 14 HR-HPV types as the cobas test. However, unlike the cobas test, the Anyplex assay is an expanded genotyping test having the ability to individually identify each of the 14 HR-HPV types simultaneously in a single analysis. The Anyplex assay has been validated in many studies elsewhere and reported as showing consistently high concordance and clinical performance with approved comparators tests (22,23). We used the Anyplex assay in the present study to test the residual cervical specimens in ThinPrep PreservCyt from our LSIL triage study that had been archived pending further genotype-specific analysis. Since the study specimens had been in storage at ambient temperature for 31–47 months following the initial cobas HPV testing of fresh specimens at the time of collection at patient enrolment, we first evaluated the integrity and suitability of the specimens for testing with the Anyplex assay by comparing test agreements and performance characteristics with those initially obtained with the cobas HPV test. We also carried out an inter-laboratory reproducibility study in verifying the suitability of the specimens. These also provided data for a secondary objective to validate the use of archived cervical specimens and the Anyplex assay.

2. Methods

2.1 Study description

The details of our initial CINtec PLUS/HPV-LSIL triage study were described in previous publications [1, 14]. Briefly, the study population comprised of patients with a history of LSIL cytology referred to the colposcopy clinic at Juravinski Hospital, Hamilton, Canada. They were enrolled over a 16-month period from November 2017 to February 2019. All study patients were attended to per standard of care, with cervical specimens collected for routine cytology in ThinPrep PreservCyt, and colposcopy and biopsies performed as indicated per routine clinical practice at enrollment. CINtec PLUS and cobas HPV tests were performed once at enrollment (baseline) using the leftover cervical cytology specimens for the study purpose. Patients’ baseline data were recorded, and the study cohort remaining under care in the colposcopy clinic was followed up to 2 years to determine disease outcome, with biopsy confirmed $\geqslant$ CIN2 serving as the clinical endpoint, and this was correlated with the baseline CINtec PLUS and cobas test results. After the completion of CINtec PLUS and HPV testing at baseline, coded residual cervical specimens were stored at ambient temperature in a climate-controlled research laboratory at St. Joseph’s Healthcare, Hamilton awaiting further genotype-specific analysis. These specimens were tested in the present study to determine genotype-specific distribution and attributable risk of HR-HPV in CIN using the Anyplex assay.

2.2 Ethics

The present study was a continuation of our CINtec PLUS/HPV-LSIL triage study. This was approved by the Hamilton Integrated Research Ethics Board and Newfoundland and Labrador Health Research Ethics Board.

2.3 cobas HPV testing

The cobas HPV test was performed at baseline in batches, using freshly collected cervical specimens no later than 6 weeks after the collection, on the Roche 4800 automated platform per manufacturer’s instructions. Results were reported as positive for genotypes 16 and or 18, and or 12 OHR types, or negative for 14 HR-HPV types.

2.4 Anyplex II HPV HR testing

Aliquots of anonymized cervical specimens under code were independently tested with the Anyplex test per manufacturer’s instructions. Laboratory personnel were blinded to the baseline cobas HPV test results and other study data. The results were reported either as positive for any of the 14 HR-HPV types individually or negative, with test failures, if any, reported as invalid. These results were verified and recorded in the study database. Anyplex testing was performed in batches over a period of one month in October 2021 at Epitome Laboratory, Kitchener, Canada. The interval between the initial baseline cobas HPV testing of fresh specimens and the subsequent Anyplex testing ranged from 31 to 47 months.

2.5 Anyplex II HPV HR inter-laboratory reproducibility study

After the completion of Anyplex testing at Epitome Laboratory, aliquots of random anonymized specimens representing $\sim$ 20% of the total were retrieved for an inter-laboratory reproducibility study. Anyplex test was performed under code independently per manufacturer’s instructions in May 2022 at Avalon Laboratories, St. John’s, Canada, and the results were verified and entered into the data base. This study was performed 7 months after testing aliquots of the same specimens at Epitome laboratory.

2.6 Analysis of Anyplex II HPV HR and cobas tests results

Data analysis was performed using SPSS for Windows, versions 23 and 27, Excel, Microsoft Office Professional Plus, 2013, MedCalc, 2021, and Social Science Statistics website, 2020. Qualitative variables were studied through different frequencies. Descriptive statistics were prepared for test results and HPV genotypes. Study data were analyzed using contingency tables to determine test positivity rates, and the diagnostic indices of cobas and Anyplex testing. All tests were two-tailed, and $p<$ 0.05 was considered statistically significant.

2.7 HPV genotype analysis

The distribution of HPV genotypes was determined among those diagnosed as having $\leqslant$ CIN1 and $\geqslant$ CIN2. The prevalence of an HPV genotype was calculated as the proportion of HPV positive samples within the above CIN grades according to cobas or Anyplex test overall and by infection type. Infection type was classified as single if only one type of HPV genotype was detected, or as multiple if more than one type detected.

Attribution of individual HPV genotypes was determined by CIN grades, CIN1, CIN2 and CIN3 including a case of adenocarcinoma in situ (AIS). To better estimate the attribution of specific genotypes to cervical lesions, we assumed a fractional allocation for each individual HPV genotype and derived apportionment for multiple type infections based on the relative frequencies observed, using a method previously described (24). To be specific, we took the relative frequencies of HPV genotypes observed as a single-type infection in lesions of a given grade and then allocated the multiple infections fractionally for each individual HPV genotype based on the relative frequencies. Differences between rates of detection of HPV genotypes were tested using Chi-square or Fisher’s exact tests, as appropriate. $p$ values of $<$ 0.05 were considered statistically significant. Statistical analyses were performed with SAS software, version 9 (SAS Institute Inc).

3. Results

3.1 Study population

The initial CINtec PLUS/HPV LISL triage study had an evaluable sample size of 598 patients [1], and of these, sufficient sample was available for 543 for the present investigation. Of these, 10 (1.8%) samples yielded invalid Anyplex results, and these could not be repeat tested due to lack of specimen and were excluded, leaving valid Anyplex test results for 533 patients (median age 33; range 21–76 years) for the study analysis. All 533 samples in the study panel had complete basic dataset across all parameters of the LSIL triage study.

3.2 Prevalence of HR-HPV and agreement of Anyplex II HPV HR test results with cobas HPV test results

In the first analysis, results from the Anyplex test were compared with the cobas test results to determine agreement and the overall HR-HPV prevalence in the total LSIL study population. Anyplex was positive in 63.0% (336/533) compared with 55.7% (297/533) by the cobas test with an overall agreement of 84.8% (Table 1). This showed a higher analytical sensitivity of Anyplex test and accordingly, the positivity rate of HR-HPV was found to be higher with Anyplex compared to cobas test. Conversely, this also indicated 37% and 44.3% of the LSIL referral population tested negative for HR-HPV by the respective tests.

Table 1
Agreement between Anyplex II HPV HR and cobas HPV tests

Test/result		cobas HPV		Total
		Positive	Negative
Anyplex II	Positive	276	60	336
HPV HR	Negative	21	176	197
Total		297	236	533

Positive agreement	276/297 $=$ 92.9%
Negative agreement	176/236 $=$ 74.6%
Overall agreement	452/533 $=$ 84.8%
Kappa, 0.69

3.3 Inter-laboratory reproducibility study

This study was conducted with a panel of 101 random specimens that had been previously tested at Epitome Laboratory. These were tested independently using the Anyplex test at Avalon Laboratories. Evaluable results were available for 94, and these are summarized in Table 2. This showed an overall agreement of 94.7% with a kappa value of 0.89, indicating a high reproducibility of the Anyplex test.

Table 2
Anyplex II HR test: Agreement of results between Epitome Laboratory and Avalon Laboratories

Test/result		Anyplex II HPV HR Epitome Laboratory		Total
		Positive	Negative
Anyplex II HPV HR	Positive	48	0	48
Avalon Laboratories	Negative	5	41	46
Total		53	41	94

Positive agreement	48/53 $=$ 90.6%
Negative agreement	41/41 $=$ 100%
Overall agreement	89/94 $=$ 94.7%
Kappa, 0.89

3.4 Performance of Anyplex II HPV HR compared to cobas HPV to detect

\geqslant

CIN2

This was based on a total of 318 of 533 patients undergoing cervical biopsy as part of their routine care, of which 88 (27.7%) were diagnosed as having $\geqslant$ CIN2/AIS, and this included 39 CIN3 and one case of AIS. For this purpose, the focus was the relative performance of the Anyplex test in detecting 14 HR types collectively in $\geqslant$ CIN2/AIS in comparison with the cobas test. As shown in Table 3, Anyplex test showed identical sensitivity to that of cobas test in detecting $\geqslant$ CIN2/AIS at 93.2% (82/88) and CIN3/AIS at 92.5% (37/40). Although Anyplex showed a slightly lower specificity than cobas test, this was not statistically significant, and there were no significant differences in positive and negative predictive values between the two tests.

Table 3
Diagnostic Indices of Anyplex II HPV HR test compared with cobas HPV test for detecting $\geqslant$ CIN2/AIS ${}^{1}$ and CIN3/AIS ${}^{1}$

$\geqslant$ CIN2/AIS ${}^{1}$		Histology			Sensitivity	Specificity	PPV	NPV
		$\geqslant$ CIN2/AIS ${}^{1}$	$\leqslant$ CIN1	Total	(95% CI)	(95% CI)	(95% CI)	(95% CI)
Anyplex II	Positive	82	154	236	93.2 (85.8–97.5)	33.0 (27.0–39.5)	34.7 (32.4–37.2)	92.7 (85.1–96.6)
HPV HR	Negative	6	76	82
	Total	88	230	318
cobas HPV	Positive	82	145	227	93.2 (85.8-97.5)	37.0 (30.7-43.6)	36.1 (33.5–38.8)	93.4 (86.5–96.9)
	Negative	6	85	91
	Total	88	230	318

CIN3/AIS ${}^{1}$		Histology			Sensitivity	Specificityx	PPV	NPV
		CIN3/AIS ${}^{1}$	$\leqslant$ CIN2	Total	(95% CI)	(95% CI)	(95% CI)	(95% CI)
Anyplex II	Positive	37	199	236	92.5 (79.6–98.4)	28.4 (23.2–34.1)	15.7 (14.2–17.3)	96.3 (89.7–98.8)
HPV HR	Negative	3	79	82
	Total	40	278	318
cobas HPV	Positive	37	190	227	92.5 (79.6–98.4)	31.7 (26.2–37.5)	16.3 (14.7–18.0)	96.7 (90.7–98.9)
	Negative	3	88	91
	Total	40	278	318

Based on 318 patients in all ages with biopsy. $\geqslant$ CIN2, cervical intraepithelial neoplasia grade 2 or worse. CIN3, cervical intraepithelial neoplasia grade 3. AIS, adenocarcinoma in situ. ${}^{1}$ Includes one case of AIS.

Table 4

Distribution of HPV genotypes in cervical intraepithelial neoplasia as detected by Anyplex II HPV HR test compared with cobas test

HPV test/genotypes		Number of samples positive for HPV
		$\leqslant$ CIN1 ( $n=$ 230)			$\geqslant$ CIN2/AIS ( $n=$ 88) ${}^{1}$
		Single ( $n=$ 130)	Multiple ( $n=$ 15)	Total ( $n=$ 230)	Single ( $n=$ 67)	Multiple ( $n=$ 15)	Total ( $n=$ 88)
cobas HPV	16 (%)	12 (9.2)	12 (80.0)	24 (10.4)	21 (31.3)	11 (73.3)	32 (36.4)
	18 (%)	3 (2.3)	4 (26.7)	7 (3.0)	2 (3.0)	6 (40.0)	8 (9.1)
	OHR (%)	115 (88.5)	14 (93.3)	129 (56.1)	44 (65.7)	15 (100)	59 (67.0)
		Single ( $n=$ 106)	Multiple ( $n=$ 48)	Total ( $n=$ 230)	Single ( $n=$ 50)	Multiple ( $n=$ 32)	Total ( $n=$ 88)
Anyplex II	16 (%)	13 (12.3)	16 (33.3)	29 (12.6)	20 (40.0)	12 (37.5)	32 (36.4)
HPV HR	18 (%)	3 (2.8)	4 (8.3)	7 (3.0)	2 (4.0)	6 (18.8)	8 (9.1)
	31 (%)	10 (9.4)	10 (20.8)	20 (8.7)	4 (8.0)	7 (21.9)	11 (12.5)
	33 (%)	4 (3.8)	1 (2.1)	5 (2.2)	1 (2.0)	2 (6.3)	3 (3.4)
	35 (%)	3 (2.8)	7 (14.6)	10 (4.3)	4 (8.0)	6 (18.8)	10 (11.4)
	39 (%)	8 (7.5)	7 (14.6)	15 (6.5)	1 (2.0)	6 (18.8)	7 (8.0)
	45 (%)	5 (4.7)	5 (10.4)	10 (4.3)	1 (2.0)	6 (18.8)	7 (8.0)
	51 (%)	10 (9.4)	13 (27.1)	23 (10.0)	6 (12.0)	8 (25.0)	14 (15.9)
	52 (%)	8 (7.5)	11 (22.9)	19 (8.3)	1 (2.0)	3 (9.4)	4 (4.5)
	56 (%)	15 (14.2)	7 (14.6)	22 (9.6)	3 (6.0)	5 (15.6)	8 (9.1)
	58 (%)	6 (5.7)	10 (20.8)	16 (7.0)	2 (4.0)	4 (12.5)	6 (6.8)
	59 (%)	7 (6.6)	9 (18.8)	16 (7.0)	1 (2.0)	3 (9.4)	4 (4.5)
	66 (%)	9 (8.5)	7 (14.6)	16 (7.0)	3 (6.0)	3 (9.4)	6 (6.8)
	68 (%)	5 (4.7)	8 (16.7)	13 (5.7)	1 (2.0)	5 (15.6)	6 (6.8)

$\leqslant$ CIN1, cervical intraepithelial neoplasia grade 1 or better. $\geqslant$ CIN2, cervical intraepithelial neoplasia grade 2 or worse. AIS, adenocarcinoma in situ. ${}^{1}$ Includes one case of AIS.

3.5 Distribution of HPV genotypes in CIN as detected by the Anyplex II HPV HR test

In this series, we analysed the distribution of HPV genotypes among those diagnosed as having $\leqslant$ CIN1 and $\geqslant$ CIN2. The prevalence of an HPV genotype was calculated as the proportion of HPV positive samples within given CIN grades, i.e., $\leqslant$ CIN1 vs. $\geqslant$ CIN2, according to Anyplex and cobas test results overall and by infection type. Infection type was classified as single if only one type of HPV genotype was detected, or as multiple if more than one type were detected.

Table 5
Attributable rate of detection of 14 HPV genotypes in cervical intraepithelial neoplasia ${}^{1}$

HPV genotypes	Number of samples positive for HPV (%)
	CIN1 ( $n=$ 68)	CIN2 ( $n=$ 48)	CIN3/AIS ( $n=$ 40) ${}^{2}$
16 (%)	9 (13.2)	13 (27.1)	16 (40.0)
18 (%)		0	4 (10.0)
31 (%)	3 (4.4)	6 (12.5)	1 (2.5)
33 (%)	0	1 (2.1)	1 (2.5)
35 (%)	3 (4.4)	4 (8.3)	4 (10.0)
39 (%)	2 (2.9)	1 (2.1)
45 (%)	3 (4.4)	3 (6.3)
51 (%)	6 (8.8)	7 (14.6)	4 (10.0)
52 (%)	6 (8.8)	0	1 (2.5)
56 (%)	4 (5.9)	3 (6.3)	1 (2.5)
58 (%)	6 (8.8)	0	4 (10.0)
59 (%)	2 (2.9)	2 (4.2)
66 (%)	5 (7.4)	3 (6.3)	2 (5.0)
68 (%)	4 (5.9)	2 (4.2)

${}^{1}$ Based on Anyplex II HPV HR test. CIN1, cervical intraepithelial neoplasia grade 1. CIN2, CIN grade 2. CIN3, CIN grade 3. AIS, adenocarcinoma in situ. ${}^{2}$ Includes one case of AIS.

Table 4 summarizes distribution of 14 HR-HPV genotypes in $\leqslant$ CIN1 vs. $\geqslant$ CIN2 by Anyplex test in comparison with cobas test in 318 of the 533 (59.7%) patients who underwent cervical biopsy as part of their routine care. Based on cobas testing, and counting single and multiple infections in $\leqslant$ CIN1, genotype 16 accounted for 10.4%, type 18 at 3.0%, and OHR at 56.1%. In contrast, in $\geqslant$ CIN2, genotype 16 accounted for 36.4%, type 18 at 9.1%, and OHR at 67.0%. Based on Anyplex testing, and counting single and multiple infections in $\leqslant$ CIN1, genotype 16 was detected in 12.6% and type 18 at 3.0%. In $\geqslant$ CIN2, these figures were, 36.4% and 9.1%, respectively. The above figures for types 16 and 18 in $\leqslant$ CIN1 were similar between cobas and Anyplex tests while these were identical in $\geqslant$ CIN2. However, the proportions testing positive for types 16 and 18 in $\leqslant$ CIN1 vs. $\geqslant$ CIN2 were significantly different ( $p<$ 0.05)

Analysing 12 OHR genotypes detected by Anyplex test in $\leqslant$ CIN1, OHR types collectively represented 80.4% (185/230) of the 14 genotypes identified (Table 4) This proportion was significantly higher than the 56.1% (129/230) collectively detected by the cobas test in $\leqslant$ CIN1 ( $p<$ 0.0001). The OHR types were often found in multiple infections, sometimes exceeding 6 types. In terms of specific OHR types identified by Anyplex testing in $\leqslant$ CIN1, types 51 and 56 accounted for $\sim$ 10% each, followed by types 31 (8.7%), 52 (8.3%) and 39, 58, 59, and 66 each accounting for $\sim$ 7%. The remaining 4 genotypes accounted for 5.7% or less each. In analysing 12 OHR genotypes detected by Anyplex test in $\geqslant$ CIN2, OHR types collectively accounted for 97.7% (86/88) vs. 67.0% (59/88) by cobas test, and this difference in the proportions testing positive for OHR types between the two tests was significant ( $p<$ 0.001). As in $\leqslant$ CIN1, these were predominantly found in multiple infections. In terms of specific OHR types identified by Anyplex test in $\geqslant$ CIN2, genotype 51 accounted for 15.9%, followed by type 31 (12.5%), type 35 (11.4%), type 56 (9.1%), and types 39 and 45 each accounting for 8.0%; these six were the most frequent types collectively accounting for 57 of 88 (64.8%) $\geqslant$ CIN2. Regardless, all 12 OHR types were detected either as a single type or in combination with other HR types in both $\leqslant$ CIN1 and $\geqslant$ CIN2 (Table 4).

3.6 Attributable rate of detection of 14 HR-HPV genotypes in CIN

In this series, the attributable rate of detection of 14 HR genotypes was determined in CIN1, CIN2 and CIN3 including a case of AIS based on the Anyplex HPV HR test. The study population comprised a total of 156 women (median age 30 years; range 21-71 years), of whom 68 had CIN1(median age 34 years, range 21–71 years), 48 had CIN2 (median age 30 years, range 21–63 years), 39 had CIN3 (median age 29 years, range 22–69 years), and one was a case of AIS.

The estimated percent attributable rate of detection for each of the 14 HR genotypes stratified by CIN/AIS grades is shown in Table 5. In CIN1, after adjustment for multiple-type infections, the most frequent genotypes in decreasing frequencies were 16, 51, 52, 58, 66, 56, 68, 31, 35, 45, 39, 59, accounting for 77.9% (53/68). In CIN2, the genotypes represented in decreasing frequencies were 16 followed by 51, 31, 35, 45, 56, 66, 59, 68, 33, 39, accounting for 93.8% (45/48). In CIN3/AIS, these were, 16 followed by types 18, 35, 51, 58, 66, 31, 33, 52, 56, accounting for 95% (38/40). After adjusting for multiple-type infections, genotype 16 showed increasing percent attribution, from 13.2% in CIN1 to 27.1% in CIN2 and 40.% in CIN3/ AIS ( $p=$ 0.0015) The single case of AIS was positive for genotype 18. Estimated percent attribution was evident for all 14 HR types across the CIN/AIS grades with all $\geqslant$ CIN2 lesions containing at least one HR type. Six genotypes, 16, 31, 35, 51, 56, and 66, were present in all grades of CIN with genotype 18 found only in CIN 3 and AIS.

4. Discussion

This investigation was conducted as part of our CINtec PLUS/HPV-LSIL triage study to determine genotype-specific distribution and attributable risk of HR-HPV in CIN grades among an LSIL referral population. This knowledge would be desirable in HPV-based screening and triage which could improve patient care and follow-up strategies. In this study, we individually characterized 14 HR-HPV genotypes, 12 of which were collectively identified as OHR types by cobas test [1, 14], using the expanded genotype-specific Anyplex test. This provided data on the overall distribution of HR-HPV genotypes and more specifically their attributable risk in biopsy-confirmed CIN grades.

There was considerable delay to getting the present study conducted, and therefore, we first established the stability of specimens over time and their suitability, and the validity of Anyplex results by comparing test agreement and clinical performance data with cobas results obtained with fresh specimens collected at enrolment. Anyplex showed a good overall agreement of 84.8% (kappa, 0.69) with cobas results (Table 1). Anyplex test has been reported to show an agreement of as high as 98% (kappa, 0.96) with the cobas test [25], and the difference could be attributable to the archived nature of specimens in our study. However, the discrepant Anyplex results in our study were actually due to a higher number of specimens testing positive by Anyplex than cobas. This implied that Anyplex test likely has a higher analytical sensitivity than cobas test, but this did not translate into higher clinical sensitivity as noted below. In fact, this could potentially impact its clinical specificity, especially in large scale screening. Regardless, Anyplex test was found to be highly reproducible with an overall agreement of 94.7% (kappa, 0.89) in the inter-laboratory study (Table 2). This is consistent with previous studies reviewed by Biazin [23]. More importantly, we analyzed diagnostic indices of Anyplex test in comparison with cobas test to further validate the clinical performance of Anyplex test and the suitability of the archived specimens in the present study. This showed a remarkable consistency and accuracy with Anyplex showing identical clinical sensitivity to that of cobas test to detect $\geqslant$ CIN2, CIN3/AIS with similar performance of other diagnostic indices (Table 3). All of the above data provided sufficient evidence on the stability of the specimens and their suitability for testing with Anyplex assay for the present study.

As previously observed in our LSIL triage study [1, 14], a large proportion of the LSIL referral population tested negative for HR-HPV. This may be due to spontaneous viral clearance during the interval ( $\sim$ 7.9 months) between the index LSIL cytology and the first colposcopy clinic visit when the HPV test was performed. This has implications in the management of LSIL referral patients in such settings in that if HPV triage is utilized, those testing negative can be safely returned to routine screening, and this could improve patient care and efficiency as articulated in our previous report [1].

In terms of distribution of HPV genotypes, since we already had genotype data based on the cobas test for types 16/18 specifically, and for OHR types collectively, this allowed us to compare these data to those obtained with Anyplex test. Consistent with previous studies [26, 27], multiple-type infections were observed more frequently than single-type infections from any grade of CIN. Counting single and multiple-type infections as detected by Anyplex in $\leqslant$ CIN1 genotype 16 accounted for 12.6% followed by type 18 at 3%. These two genotypes are known to cause persistent infection and thus are significantly more associated with lesion progression than OHR types, and this was evident in $\geqslant$ CIN2, with proportions of types 16 and 18 rising to 36.4% and 9.1%, respectively. The above differences observed in the association of genotypes 16 and 18 in $\leqslant$ CIN1 vs. $\geqslant$ CIN2 were significant and similar between the two tests, showing an excellent agreement between cobas and Anyplex tests in detecting genotypes 16 and 18 and attesting to the reliability of both tests. However, by comparing the data of Anyplex and cobas tests for detection of the 12 OHR types, we observed significant differences between the two tests in the proportions testing positive in both CIN categories i.e., 80.4% by Anyplex and vs. 56.1% by cobas in $\leqslant$ CIN1; and 97.7% vs. 67%, respectively, in $\geqslant$ CIN2. This can be explained by the fact the above figures were based on mostly multiple-type counts, and Anyplex test having a higher analytical sensitivity detects multiple OHR types with greater precision and at higher frequency than cobas test.

In terms of specific OHR types identified by Anyplex test in $\leqslant$ CIN1, our study indicated, in deceasing order, types 51, 56, 31, 52, 58, 59, 66, 39, and 68 were more frequently detected than the other 3 types, 33, 35, and 45. In $\geqslant$ CIN2, types 51, 31, 35, 56, 39 and 45 were more frequently detected than the other 6 types, 33, 52, 58, 59, 66 and 68. Overall, 4 genotypes, 31, 35, 45 and 51 were found to be more associated with $\geqslant$ CIN2 than $\leqslant$ CIN1 compared to the remaining 8 OHR types (Table 4). The above data in CIN categories provide some indication of the relative prevalence of OHR types, but the association of specific types in $\geqslant$ CIN2 likely implies persistence and potential lesion progression, and this has clinical implications. Regardless, it’s important to note that all 12 OHR types were detected, often in multiple-type infections, in both $\leqslant$ CIN1 and $\geqslant$ CIN2 as previously reported [26].

We assessed the attributable rate of detection of all 14 HR genotypes by CIN grades including a single case of AIS after adjustment for multiple-type infections. In CIN1, the most frequent genotypes in decreasing order were 16, 51, 52, 58, 66, 56, 68, 31, 35, 45, 39, 59 accounting for 77.9%. In CIN2, these were 16, 51, 31, 35, 45, 56, 66, 59, 68, 33, 39, accounting for 93.8%. In CIN3/AIS, these were 16, 18, 35, 51, 58, 66 ,31, 33, 52, 56 accounting for 95.0%. Of these, only type 16 showed a significantly increasing percent attribution, from 13.2% in CIN1 to 27.1% in CIN2 and 40.0% in CIN3/AIS (Table 5). This is consistent with the fact that persistent transforming infections are mostly associated with genotype 16 than with any other type as shown in numerous studies [15, 16, 27], and underscores the importance of detection of this genotype as well as the genotype-specific risk threshold recommended in HPV primary screening [20]. Besides genotype 16, detection of types 31, 35, 51, 56 and 66 in all grades of CIN would appear to indicate their relative importance in causing persistent infection and lesion progression compared to the rest of the OHR types. However, estimated percent attribution was evident for all 14 HR types across the CIN/AIS grades with all $\geqslant$ CIN 2 lesions containing at least one HR type, and this highlights the importance of considering all 14 HR-HPV types in assessing cervical cancer risk. Its worth noting that currently marketed commercial HPV tests from major manufacturers target the 14 HR-HPV types, some of them with features to simultaneously identify specific genotypes individually or in pools [1, 18, 19, 23, 28]. This compliments the presently available second-generation nonavalent HPV vaccine against the most prevalent HR genotypes, having the potential to prevent an additional $\sim$ 20% of cervical cancers attributable to non-16 and 18 genotypes [29].

Our study data on genotype distribution are similar to our previous study in a Canadian population [26] and other large-scale studies elsewhere. However, our study was based on a small sample size with a limited number of $\geqslant$ CIN2 cases, representing a single region in the country. Further, although our LSIL-triage population comprised exclusively of women with LSIL referred to colposcopy, at the time of their first visit to the colposcopy clinic and their enrolment at that point into the study, only 41.5% were found to still have LSIL with 48.9% showing regression, implying viral clearance in a significant proportion during the interval between index referral LSIL cytology and enrolment [14]. Therefore, it should be noted the genotype data we generated was not based on a concurrent LSIL population. Moreover, types 18 and 45 were underrepresented in our study considering their importance in invasive cervical cancer. Further, we used a method to correct for multiple-type infections that assumes fractional allocation for each individual type. This modified the frequency order for several types and reduced the frequency of HPV types in CIN grades compared with crude estimates of prevalence, but it likely represents the actual attribution.

Besides the genotype data we generated using the Anyplex test, this study also offered an opportunity to assess the clinical performance of Anyplex test in comparison with the cobas test. While Anyplex has been evaluated in many studies elsewhere with superior sensitivity and high levels of concordance with other validated tests [22, 23], ours is the first clinical assessment study of the Anyplex test in a Canadian population. This test performed non-inferior to cobas test with identical sensitivity to detect $\geqslant$ CIN2 and can serve as a suitable test in HPV primary cervical screening and cytology triage. Unlike many other current generations of HPV test, the Anyplex test offers the advantage of specifically identifying individual HR genotypes and this may be useful in clinical practice including assessing genotype-specific persistence and molecular epidemiological studies. It should be noted that the reduced positivity rate OHR with cobas test may be attributed to limited detection of multiple types. However, this did not affect the clinical efficacy of cobas test in identifying $\geqslant$ CIN2/AIS.

This study was conducted with cervical specimens collected in PreservCyt that had been stored in ambient temperature for prolonged time. Our study data attested to the stability of specimens in PreservCyt for an extended time for HPV testing, and the ability of DNA extraction and detection methods of the Anyplex test in performing the test. However, our study showed invalid results in 1.8% of the specimens tested. In this regard, we note a study of the Anyplex test conducted within the VALGENT framework which used archived PreservCyt specimens stored at 70 ${}^{\circ}$ C found invalid results only in 0.1% (2/1600) [30]. This difference is most likely due to degradation of DNA in our study specimens at ambient temperature over time. In an earlier study, Castle et al [31] reported approximately 15% of PreservCyt specimens could not be amplified for any $\beta$ -globin DNA fragment after 5 years of storage at ambient temperatures, indicating cervical specimens archived in PreservCyt underwent partial DNA degradation after several years of storage in ambient temperatures. A more recent study of cervical specimens collected in PreservCyt reported HPV DNA was stable when specimens were stored at 2–8 ${}^{\circ}$ C for more than 2.5 years [32]. Our data and those of the others cited above provide information on prolonged storage temperatures and stability of cervical specimens in PresevCyt for detection of HPV DNA.

5. Conclusions

Among LSIL referral population, genotypes 16/18 were far more prevalent in $\geqslant$ CIN2 than $\leqslant$ CIN1 and this underscores their importance in lesion progression with persistent infection and a significant attributable risk for cervical cancer especially with genotype 16. While the 12 OHR types were found widely distributed in both $\leqslant$ CIN1 and $\geqslant$ CIN2, the attributable risk was evident for each of these genotypes in $\geqslant$ CIN2. Testing for these genotypes in HPV primary cervical screening and cytology triage could identify those at increased risk of cervical cancer and also be beneficial in the management of LSIL referral populations. Further, Anyplex test performed identical to that of cobas test in detecting $\geqslant$ CIN2 with an added advantage of genotype-specific results for each of the 14 HR-HPV.

Authors’ contributions

Conception: SR, DJ.

Interpretation or analysis of data: SR, DJ, RA, LG, YX, WW, PA, AG, DJS, MC.

Preparation of the manuscript: SR.

Revision for important intellectual content: SR, RA, MC.

Supervision: SR, DJ.

All authors reviewed and approved the final version for submission. All authors attest they meet the ICMJE criteria for authorship.

Conflict of interest disclosures

Sam Ratnam received research grant from Seegene Inc., Seoul, South Korea. He also received research grant unrelated to this study from Roche Diagnostics, Montreal, Canada. Max Chernesky received research grant unrelated to this study from Roche Diagnostics. Other members of the study team declare no conflict of interests.

Funding support

This study was supported by a research grant from Seegene Inc., Seoul, South Korea. The study was investigator initiated and conducted independently. Seegene Inc., offered suggestions with the study design and reviewed the manuscript, but did not have any input or influence in the conduct or reporting the results of this study.

Footnotes

Acknowledgments

We thank James Yantzi, Seegene Inc., Canada, Toronto, Canada for his support, and Sung Hee Park and Jinny Ha, Seegene Inc., Seoul, South Korea for the provision of test kits, funding support and valuable suggestions. We also thank Teri-Lynn Steeves, Epitome Genetics, Kitchener, Canada for technical support, Dr. Dustin Costescu, Dr. Miranda Schell, Anne Ecobichon-Morris, Mary Jane Carrier, and other colposcopy clinic medical and nursing staff, Juravinski Hospital, Hamilton, Canada for assistance with patient enrolment and follow-up for the initial CINtec PLUS-HPV triage study; Rob Needle, Eastern Health, St. John’s, Canada for cobas HPV testing, and Holly Edwards and Tania Smith, Eastern Health, St. John’s, Canada for administrative assistance. Our special thanks to Ralph Insinga, Merck Research Laboratories, North Wales, Pennsylvania, USA for his advice and help with genotype analysis.

References

Gilbert

Ratnam

Jang

et al., Comparison of CINtec PLUS cytology and cobas HPV test for triaging Canadian patients with LSIL cytology referred to colposcopy: A two-year prospective study, Cancer Biomark 34 (2022), 347–358.

Wright

T.C.

Jr., and Schiffman

, Adding a test for human papillomavirus DNA to cervical-cancer screening, N Engl J Med 348 (2003), 489–90. doi: 10.1056/NEJMp020178.

Doorbar

Quint

Banks

et al. The biology and life-cycle of human papillomaviruses, Vaccine 30S (2012), F55–70.

Dillner

Rebolj

Birembaut

et al., Long term predictive values of cytology and human papillomavirus testing in cervical cancer screening: joint European cohort study, BMJ 337 (2008), a1754.

Kitchener

H.C.

Gilham

Sargent

et al., A comparison of HPV DNA testing and liquid based cytology over three rounds of primary cervical screening: extended follow up in the ARTISTIC trial, Eur J Canc. 47 (2011), 864–871.

Arbyn

Ronco

Anttila

et al. Evidence regarding human papillomavirus testing in secondary prevention of cervical cancer, Vaccine 30S (2012), F88–F99.

Ronco

Dillner

Elfstrom

K.M.

et al., Efficacy of HPV-based screening for prevention of invasive cervical cancer: followup of four European randomised controlled trials, Lancet 383 (2014), 524–532.

Wright

T.C.

Jr. Cox

J.T.

Massad

L.S.

et al., Conference, Consensus Guidelines for the management of women with cervical cytological abnormalities, J Amer Med Assoc 287 (2001), 2120–2129.

Solomon

Schiffman

and Tarone

, ALTS Study group. Comparison of three management strategies for patients with atypical squamous cells of undetermined significance: baseline results from a randomized trial, J Natl Cancer Inst 93 (2001), 293–299.

10.

Wright

T.C.

Stoler

M.H.

Behrens

C.M.

et al., Primary cervical cancer screening with human papillomavirus: end of study results from the ATHENA study using HPV as the first-line screening test, Gynecol Oncol 136 (2015), 189–197.

11.

Fontham

E.T.H.

Wolf

A.M.D.

Church

T.R.

et al., Cervical cancer screening for individuals at average risk: 2020 guideline update from the American Cancer Society, CA Cancer J Clin 70 (2020), 321–346.

12.

Ronco

Cuzick

Segnan

et al., HPV triage for low grade (L-SIL) cytology is appropriate for women over 35 in mass cervical cancer screening using liquid-based cytology, Eur J Cancer 43 (2007), 476–480.

13.

Cuzick

Cox

J.T.

Zhang

et al., Human papillomavirus testing for triage of women with low-grade squamous intraepithelial lesions, Int J Cancer. 132 (2013), 959–966.

14.

Ratnam

Jang

Gilbert

et al., CINtec PLUS and cobas HPV testing for triaging Canadian women referred to colposcopy with a history of low grade-squamous intraepithelial lesion: Baseline findings, Papillomavirus Res 10 (2020), 100206.

15.

Munoz

Bosch

F.X.

de Sanjose

et al., Epidemiologic classification of human papillomavirus types associated with cervical cancer, N Engl J Med. 348 (2003), 518–527.

16.

Khan

M.J.

Castle

P.E.

Lorincz

A.T.

et al., The elevated 10-year risk of cervical precancer and cancer in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of type-specific HPV testing in clinical practice, J Natl Cancer Inst 97 (2005), 1072–1079.

17.

de Sanjose

Quint

W.G.

Alemany

et al., Human papillomavirus genotype attribution in invasive cervical cancer: a retrospective cross-sectional worldwide study, Lancet Oncol 11 (2010), 1048–1056.

18.

Poljak

Ostrbenk

Seme

et al., Three year longitudinal data on the clinical performance of the Abbott RealTime High Risk HPV test in a cervical cancer screening setting, J Clin Virol 76(Suppl 1) (2016), S29–S39.

19.

Bonde

J.H.

Pedersen

Quint

et al., Clinical and analytical performance of the BD Onclarity HPV assay with SurePath screening samples from the Danish cervical screening program using the VALGENT framework, J Clin Microbiol. (2020), 58 e01518–19.

20.

Huh

W.K.

Ault

K.A.

Chelmow

et al., Use of primary high-risk human papillomavirus testing for cervical cancer screening: interim clinical guidance, Gynecol Oncol 136 (2015), 178–182.

21.

Castellsague

, Natural history and epidemiology of HPV infection and cervical cancer, Gynecol Oncol 110(3 Suppl 2) (2008), S4–7. doi: 10.1016/jygyno.2008.07.045.

22.

Cornall

A.M.

Poljak

Garland

S.M.

et al., Anyplex II HPV28 detection and Anyplex II HPV HR assays are highy corcordant with other commercial assays for detection of high-risk HPV genotypes in women with high-grade cervical abnormalities, Eur J Clin Microbiol Infect Dis 36 (2017), 545–551.

23.

Biazin

, Corcordance of Anyplex II HPV HR assays with reference HPV assays in cervical cancer sreening: Systematic review, J Virol Methods 301 (2022), 114435.

24.

Insinga

R.P.

Liaw

K.L.

Johnson

L.G.

et al., A systematic review of the prevalence and attribution of human papillomavirus types among cervical, vaginal and vulvar precancers and cancers in the United States, Cancer Epidemiol Biomarkers Prev 17 (2008), 1611–1622.

25.

Lee

D.H.

Hwang

N.R.

Lim

M.C.

, et al., Comparison of the performance of Anyplex II HPV HR the cobas 4800 human papillomavirus test and Hybrid Capture 2, Ann Clin Biochem 53 (2016), 561–567.

26.

Coutlee

Ratnam

Ramanakumar

A.V.

, et al. Distribution of Human papillomavirus genotypes in cervical intraepithelial neoplasia and invasive cervical cancer in Canada, J Med Virology 83 (2011), 1034–1041.

27.

Clifford

Franceschi

Diaz

, et al. HPV type-distribution in women with and without cervical neoplastic diseases, Vaccine 24 (2006), S26–S34.

28.

Ratnam

Coutlee

Fontaine

et al. APTIMA HPV E6/E7 mRNA test is as sensitive as Hybrid Capture 2 assay but more specific at detecting cervical precancer and cancer, J Clin Microbiol 49 (2011), 557–564.

29.

Serrano

Alemany

Tous

, et al. Potential impact of a nine-valent vaccine in human papillomavirus related cervical disease, Infect Agent Cancer 7 (2012), 38. doi: 10.1186/1750-9378-7-38.

30.

Ostrbenk

Arbyn

, et al., Clinical and analytical evaluation of the Anyplex II HPV HR detection assay within the VALGENT framework, J Clin Microbiol 56 (2018), e01176–18.

31.

Castle

P.E.

Solomon

Hildesheim

, et al. Stability of archived liquid-based cervical cytologic specimens, Cancer 99 (2003), 89–96.

32.

Agreda

P.M.

Beitman

G.H.

Gutierrez

E.C.

, et al. Long-term stability of human genomic and human papillomavirus DNA stored in BD SurePath and Hologic PreservCyt liquid-based cytology media, J Clin Microbiol 51 (2013), 2702–2706.

HPV genotype-specific distribution and attributable risk in cervical intraepithelial neoplasia in a referral population with a history of LSIL

Abstract

BACKGROUND AND OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Methods

2.1 Study description

2.2 Ethics

2.3 cobas HPV testing

2.4 Anyplex II HPV HR testing

2.5 Anyplex II HPV HR inter-laboratory reproducibility study

2.6 Analysis of Anyplex II HPV HR and cobas tests results

2.7 HPV genotype analysis

3. Results

3.1 Study population

3.2 Prevalence of HR-HPV and agreement of Anyplex II HPV HR test results with cobas HPV test results

Table 1 Agreement between Anyplex II HPV HR and cobas HPV tests

Table 2 Anyplex II HR test: Agreement of results between Epitome Laboratory and Avalon Laboratories

Table 3 Diagnostic Indices of Anyplex II HPV HR test compared with cobas HPV test for detecting ⩾ CIN2/AIS 1 and CIN3/AIS 1

Table 5 Attributable rate of detection of 14 HPV genotypes in cervical intraepithelial neoplasia 1

4. Discussion

5. Conclusions

Authors’ contributions

Conflict of interest disclosures

Funding support

Footnotes

Acknowledgments

References

Table 1
Agreement between Anyplex II HPV HR and cobas HPV tests

Table 2
Anyplex II HR test: Agreement of results between Epitome Laboratory and Avalon Laboratories

Table 3
Diagnostic Indices of Anyplex II HPV HR test compared with cobas HPV test for detecting $\geqslant$ CIN2/AIS ${}^{1}$ and CIN3/AIS ${}^{1}$

Table 5
Attributable rate of detection of 14 HPV genotypes in cervical intraepithelial neoplasia ${}^{1}$