Agreement between colposcopic impression and histological diagnosis among human papillomavirus type 16-positive women: a clinical trial using dynamic spectral imaging colposcopy
Drs A Zaal, Division of Woman and Baby, Gynaecological Oncology, University Medical Center Utrecht, PO Box 85500, 3508 GA Utrecht, the Netherlands. Email email@example.com
Please cite this paper as: Zaal A, Louwers J, Berkhof J, Kocken M, ter Harmsel W, Graziosi G, Spruijt J, Balas C, Papagiannakis E, Snijders P, Meijer C, van Kemenade F, Verheijen R. Agreement between colposcopic impression and histological diagnosis among human papillomavirus type 16-positive women: a clinical trial using dynamic spectral imaging colposcopy. BJOG 2012;119:537–544.
Objective To investigate the agreement between conventional colposcopic impression, dynamic spectral imaging (DSI) colposcopy and histology, for human papillomavirus type 16-positive (HPV16+) and non-16 high-risk (hr) HPV+ women.
Design Prospective, comparative, multicentre clinical trial.
Setting Three colposcopy clinics in the Netherlands.
Population Women (n = 177) aged 18 years or over with an intact cervix, referred for colposcopy.
Methods The colposcopist graded the lesion by using the DSI colposcope as a regular video colposcope. Subsequently the DSI impression was displayed and biopsies were taken from all abnormal areas as well as from a random (normal) site. A cervical smear was taken for HPV typing.
Main outcome measures Histologically confirmed high-grade cervical intraepithelial neoplasia or cancer (CIN2+), positive for HPV16 or for any other hrHPV type.
Results The DSI colposcope identified more CIN2+ cervical lesions among HPV16+ women than in non-16 hrHPV+ women (P = 0.032 regardless of final histology and P = 0.009 among women with CIN2+). Consequently, the sensitivity of the DSI colposcope for detecting CIN2+ lesions was higher in HPV16+ women than in non-16 hrHPV+ women (97% versus 74%, P = 0.009). No such differences were seen for the colposcopist impression. In addition, mainly smaller cervical lesions are missed by the colposcopist.
Conclusions The sensitivity of DSI colposcopy for CIN2+ is higher in HPV16+ than in non-16 hrHPV+ women. Furthermore, regardless of HPV16 status, the sensitivity of DSI for CIN2+ is higher than that of the colposcopist, probably because colposcopists tend to miss smaller cervical lesions.
Uterine cervical cancer is the second most frequent invasive cancer in women worldwide and is caused by malignant transformation of cervical epithelium induced by high-risk human papillomavirus (hrHPV) through premalignant stages of cervical intraepithelial neoplasia (CIN).1
The incidence and mortality rates of cervical cancer in developed countries have been reduced at least partly as a result of population-based screening programmes and subsequent treatment of premalignant cervical lesions.2 The efficacy of these programmes depends strongly on the recognition of high-grade cervical abnormalities by colposcopy after referral. Unfortunately the sensitivity of colposcopy to detect high-grade cervical lesions is low, around 55%.3–6 It has previously been shown by us and others that the sensitivity of colposcopy can be significantly improved when the dynamic spectral imaging (DSI) colposcope is used.7,8
Recently in a sub-study of the ALTS trial (atypical squamous cells of undetermined significance [ASCUS]-low-grade squamous intraepithelial lesion [LSIL] Triage Study)9 Jeronimo et al.10 observed that HPV16 positive (HPV16+) lesions produce more defined visual abnormalities at colposcopy than lesions caused by other hrHPV types, regardless of final histological diagnosis. As a consequence, it has been hypothesised that HPV16+ lesions are detected earlier than other lesions.11
To further explore the relationship between hrHPV infection, colposcopic appearance and histology, we compared lesions related to various hrHPV types using the DSI colposcope as the most sensitive instrument in a clinical study population8 and evaluated the ability of DSI to correctly classify high-grade cervical lesions in HPV16+ women.
This study was designed as a sub-study of the DSI colposcope validation trial,8 a prospective multicentre comparative clinical trial in three Dutch outpatient clinics (VU University Medical Center, Amsterdam; Reinier de Graaf Hospital, Voorburg and Sint Antonius Hospital, Nieuwegein). In this trial women aged 18 years and over who were referred to the outpatient clinic for regular colposcopy were consecutively included. The indication for colposcopy was an abnormal cytological test result (twice borderline or mild dyskaryosis [BMD] or once worse than [>] BMD) or follow up of an untreated CIN1 or CIN2 lesion. BMD corresponds to ASCUS/ASC-H (cannot exclude high-grade squamous intraepithelial lesion)/LSIL and > BMD equates to high-grade squamous intraepithelial lesion.12 For this sub-study, only women with an adequate HPV test result, who were treated ‘according to protocol’, were evaluated (n = 177, 96.7% of the ‘according to protocol’ cohort). Further details on patient selection, trial design and training of the colposcopists have been reported previously.8
The institutional review boards of the three participating clinics approved the protocol and the study was registered in the Dutch trial registry (ISRCTN66112760). Signed informed consent was obtained from all women before any study procedures.
After taking a cervical scrape for HPV testing, colposcopy was performed using the DSI digital colposcope (DySIS® v2.1; DySIS Medical Ltd [formerly Forth Photonics Ltd], Livingston, UK) using a standard 3% acetic acid solution. The DSI digital colposcope enables the calculation of a colour-coded map that is based on quantifying the dynamics (temporal behaviour and intensity) of acetowhitening recorded for a 3-minute period for every location on the cervix. During the procedures in the trial, the DSI colour-coded map was displayed only after the colposcopist’s impression (most atypical site and grading) had been digitally recorded. Impression was defined and recoded in the final data analysis as normal to low-grade (N/LG) or high-grade (HG). For DSI results analysis, the colours red and above on its colour scale (cyan, blue, green, red, yellow and white) were considered as indicative for the presence of high-grade disease.
A desktop version of the DSI software was used to perform area measurements on the collected data; the standardised set-up of the colposcope (field of view and working distance) allows the direct quantification and comparison of the size of the examined area. Expert users (JAL, MK, AZ) used a graphic tool to indicate the area of the cervix on each colposcopic image and high-grade lesions were subsequently quantified in terms of pixel counts based on the DSI colour-coded map. The total examination area was 825 mm2 and equalled 786 432 pixels. In this area, 1 mm2 equals 953.25 pixels, so one pixel area is 0.001049 mm2. When translating from image pixels to mm one has to keep in mind that the DSI system calculates the pixels from a two-dimensional image, constructed from the three-dimensional surface of the cervix. Therefore the actual size in mm can differ from the size calculated by DSI, but can still be considered as a good approximation for the purpose of comparison.13
Histology was considered the ‘gold standard’ in this study. Biopsies were taken from all sites that were suspect for high-grade disease (indicated by either the colposcopist or DSI) and in addition a ‘random’ biopsy was taken from all women. This random biopsy was taken from apparently normal cervical tissue on the opposite side of the lesion(s). If both colposcopist and the DSI colposcope evaluated the cervix as normal, one biopsy was taken from the transformation zone at the 12 o’clock position, to reduce the possibility of missing a lesion. Women diagnosed with high-grade disease were treated according to current national guidelines.14
Independent histopathology review of all biopsy and treatment specimens was performed by a pathologist specialising in gynaecological pathology (FK). In case of disagreement between original assessment and review (defined as N/LG [CIN0/1] versus HG or cancer [CIN2+]) a third expert reviewer (CM) graded the lesion blinded to all previous results. The final diagnosis, defined as the most abnormal diagnosis in all specimens (biopsy, endocervical curettage, large loop excision of the transformation zone, cold-knife conisation or hysterectomy, whichever applicable) was based on a majority decision, and was classified as CIN0/1, CIN2+ or CIN3+.
HPV detection and typing
Cervical scrapes were collected in Universal Collection Medium (Qiagen Corporation, Gaithersburg, MD, USA) and DNA was extracted from the cervical smear specimens using proteinase K digestion according to standard procedures.15 Detection and genotyping of HPV was performed using the clinically validated GP5+/6+ polymerase chain reaction with an enzyme immunoassay readout followed by reverse line blot analysis16,17 of enzyme immunoassay-positive cases to identify hrHPV (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68) and low-risk (lr) HPV genotypes (6, 11, 26, 34, 40, 42, 43, 44, 53, 54, 55, 57, 61, 70, 71, 72, 73, 81, 82, 83, 84 and 89).
All clinical data collected were analysed using SPSS (software package version 15.0, Chicago, IL, USA). For all statistical tests a two-tailed P-value ≤ 0.05 was considered significant. Difference in colposcopic impression and histological outcome in women positive for HPV16 only or co-infected with any other HPV type (HPV16+) compared with women negative for HPV16 but positive for at least one other hrHPV type (non-16 hrHPV+) was calculated using two-sided Fisher’s Exact Testing. Mann–Whitney U testing was performed to assess whether the number of ‘HG’ pixels in the DSI colour-coded map (a reflection of the lesion size) was related to the HPV16 status and the ability to correctly classify a lesion as HG.
A total of 183 women were included in the ‘according to protocol’ cohort of the DSI validation study.8 Of these, six women were excluded from the current analysis because no HPV sample was taken or because of testing failure, resulting in a cohort of 177 women. The HPV test was positive in 133 (75.1%) of the 177 women, and subsequent typing resulted in lrHPV+ (10), non-16 hrHPV+ (80) or hrHPV16+ (42). In one hrHPV+ woman the type could not be determined (HPV X).
The population baseline characteristics can be seen from Table 1, stratified according to HPV testing and typing results. The majority (n = 160; 90.4%) of women were referred for colposcopy because of an abnormal cytological test result whether or not in combination with a positive hrHPV test. In all other women (n = 17; 9.6%) the indication for colposcopy was follow-up of cervical lesions. Of the 177 women included, four (2.3%) had a normal, 113 (63.8%) BMD and 60 (33.9%) had a > BMD cervical cytology referral result. Final histology was N/LG in 92 (52.0%) and HG in 85 (48.0%) women.
Table 1. Baseline characteristics and study findings for all 177 women in this study, stratified according to result of HPV typing (the total of the stratified groups is 176, because the HPV typing failed in one woman)
|Median age [range]||33.6 [18.7–62.6]||40.0 [21.4–61.6]||39.9 [23.4–45.8]||33.2 [18.6–62.5]||33.6 [19.8–50.6]|
|Mean pregnancies (SD)||1.2 (1.3)||1.5 (1.5)||0.8 (0.8)||1.1 (1.2)||1.4 (1.4)|
|Mean sex partners (SD)|
|In total||2.9 (1.1)||2.5 (1.2)||3.3 (1.0)||2.8 (1.1)||2.9 (1.1)|
|In last year||1.3 (0.9)||1.1 (0.8)||1.4 (1.3)||1.3 (0.9)||1.4 (1.0)|
|Smoking, n (%)||66 (37.3)||15 (34.1)||4 (40.0)||31 (38.8)||16 (38.1)|
|Condom use, n (%)|
|Always||9 (5.1)||0 (0)||1 (10.0)||4 (5.0)||4 (9.5)|
|Sometimes||33 (18.6)||6 (13.6)||3 (30.0)||15 (18.8)||9 (21.4)|
|Never||119 (67.2)||31 (70.5)||4 (40.0)||54 (67.5)||29 (69.0)|
|Missing/NA||16 (9.1)||7 (15.9)||2 (20.0)||7 (8.8)||0|
|Menopausal status, n (%)|
|Pre/peri||152 (85.9)||37 (84.1)||7 (70.0)||68 (85.0)||39 (92.9)|
|Post||13 (7.3)||5 (11.4)||1 (10.0)||5 (6.3)||2 (4.8)|
|Unknown||12 (6.8)||2 (4.5)||2 (20.0)||7 (8.8)||1 (2.4)|
|Indication for colposcopy, n (%)|
|Abnormal cytology||129 (72.9)||32 (72.7)||7 (70.0)||59 (73.8)||30 (71.4)|
|Abnormal cytology and HPV+||31 (17.5)||2 (4.5)||2 (20.0)||20 (25.0)||7 (16.7)|
|Follow up CIN1||15 (8.5)||8 (18.2)||1 (10.0)||1 (1.3)||5 (11.9)|
|Follow up CIN2||2 (1.1)||2 (4.5)||0||0||0|
|Last cytology, n (%)|
|Normal||4 (2.3)||2 (4.5)||0||1 (1.3)||1 (2.4)|
|BMD||113 (63.8)||28 (63.6)||7 (70.0)||57 (71.3)||21 (50.0)|
|> BMD||60 (33.9)||14 (31.8)||3 (30.0)||22 (27.5)||20 (47.6)|
|Median biopsies* [range]||2.0 [0–4]||2.0 [0–3]||2.0 [0–4]||2.0 [0–4]||2.0 [0–3]|
|Endocervical curettage, n (%)||18 (10.2)||7 (15.9)||1 (10.0)||10 (12.5)||0|
|SCJ, n (%)|
|On ectocervix||153 (86.4)||35 (79.5)||9 (90.0)||67 (83.3)||42 (100.0)|
|In endocervix||12 (6.8)||4 (9.1)||0||8 (10.0)||0|
|Visualised after spreading||11 (6.2)||4 (9.1)||1 (10.0)||5 (6.3)||0|
|Not completely visualised||1 (0.6)||1 (2.3)||0||0||0|
|Histology, n (%)|
|No CIN||43 (24.3)||20 (45.5)||3 (30.0)||16 (20.0)||4 (9.5)|
|CIN1||49 (27.7)||14 (31.8)||2 (20.0)||26 (32.5)||6 (14.3)|
|CIN2||16 (9.0)||1 (2.3)||1 (10.0)||9 (11.3)||5 (11.9)|
|CIN3||66 (37.3)||9 (20.5)||4 (40.0)||28 (35.0)||25 (59.5)|
|AdCa||3 (1.7)||0||0||1 (1.3)||2 (4.8)|
|Histology group, n (%)|
|Normal/LG||92 (52.0)||34 (77.3)||5 (50.0)||42 (52.5)||10 (23.8)|
|HG||85 (48.0)||10 (22.7)||5 (50.0)||38 (47.5)||32 (76.2)|
|Mean HPV types (SD)||NA||NA||1.4 (0.5)||1.8 (1.1)||1.6 (1.2)|
|Mean lrHPV types (SD)||NA||NA||1.4 (0.5)||0.3 (0.7)||0.3 (0.6)|
|Mean hrHPV types (SD)||NA||NA||NA||1.6 (0.9)||1.3 (0.8)|
|Multiple hrHPV types, n (%)||NA||NA||NA||24 (30.0)||9 (21.4)|
The prevalence of the individual HPV types among the 132 HPV-positive women was as follows; HPV16 (n = 42; 31.8%), HPV31 (n = 21; 15.9%), HPV51 n = 16; 12.1%), HPV52 (n = 13; 9.8%), HPV18 (n = 12; 9.1%), HPV42 (n = 12; 9.1%), HPV56 (n = 11; 8.3%), HPV66 (n = 10; 7.6%), HPV33 (n = 9; 6.8%), HPV45 (n = 9; 6.8%), HPV59 (n = 8; 6.1%), HPV6 (n = 8; 6.1%), HPV39 (n = 7; 5.3%), HPV73 (n = 7; 5.3%), HPV58 (n = 6; 4.5%), HPV54 (n = 6; 4.5%), HPV67 (n = 3; 2.3%), HPV35 (n = 2; 1.5%), HPV53 (n = 2; 1.5%). HPV type 26, 82, 11, 30, 40, 70 and 81 were each detected only once, (0.8%). Both single and multiple infections were present and the above prevalence was calculated per individual HPV type. Multiple hrHPV infections were detected in 24 (30.0%) of the non-16 hrHPV+ women and in nine (21.4%) of the HPV16+ women (Table 1).
HPV16+ versus non-16 hrHPV+ cervical lesions
Regardless of histological outcome, DSI indicated a higher percentage of lesions as being HG in HPV16+ than in non-16 hrHPV+ women (73.8% versus 52.5%, P = 0.032; Table 2). Among HPV16+ women all ten women with CIN0/1 lesions were correctly classified by DSI as N/LG, whereas among non-16 hrHPV+ women 14 (33.3%) of the 42 CIN0/1 lesions were incorrectly classified as HG (P = 0.046). Moreover, the sensitivity of DSI to detect CIN2+ lesions was higher among HPV16+ women than among non-16 hrHPV+ women (97%, 95% CI 84–100 versus 74%, 57–87; P = 0.009). For women with CIN3+ lesions sensitivity figures were similar, although not statistically significant (96%, 95% CI 81–99 versus 79%, 61–90; P = 0.103). In contrast, no significant differences in sensitivity for CIN2+ were observed for the individual colposcopist impression between HPV16+ and non-16 hrHPV+ women, either when all histology was combined, or when stratified per histological outcome (P = 0.274 to P = 1.0; Table 2).
Table 2. Colposcopic impression of DSI and the colposcopist, defined as normal or low-grade (N/LG) versus high-grade (HG) cervical intraepithelial neoplasia
|Prediction by DSI|
| N/LG||33 (75)||5 (50)||38 (47.5)||11 (26.2)||87||0.032|
| HG||11 (25)||5 (50)||42 (52.5)||31 (73.8)||89|
| N/LG||28 (82.4)||4 (80)||28 (66.7)||10 (100)||70||0.046|
| HG||6 (17.6)||1 (20)||14 (33.3)||0 (0)||21|
| N/LG||5 (50)||1 (20)||10 (26.3)||1 (3.1)||17||0.009|
| HG||5 (50)||4 (80)||28 (73.7)||31 (96.9)||68|
| N/LG||4 (44.4)||1 (25.0)||6 (20.7)||1 (3.7)||12||0.103|
| HG||5 (55.6)||3 (75.0)||23 (79.3)||26 (96.3)||57|
|Prediction by colposcopist|
| N/LG||32 (72.7)||9 (90)||50 (62.5)||24 (57.1)||115||0.566 |
| HG||12 (27.3)||1 (10)||30 (37.5)||18 (42.9)||61|
| N/LG||28 (82.4)||5 (100)||35 (83.3)||9 (90)||77||1.0 |
| HG||6 (17.6)||0 (0)||7 (16.7)||1 (10)||14|
| N/LG||4 (40)||4 (80)||15 (39.5)||15 (46.9)||38||0.630|
| HG||6 (60)||1 (20)||23 (60.5)||17 (53.1)||47|
| N/LG||3 (33.3)||3 (75.0)||9 (31.0)||13 (48.1)||28||0.274|
| HG||6 (66.7)||1 (25.0)||20 (69)||14 (51.9)||41|
The sensitivity and specificity with 95% CI for lesion detection at the threshold of CIN2+ was calculated for the total study population and specified according to HPV16 status (Table 3). Although not statistically significant, there was a trend towards more DSI-guided biopsies from HPV16+ than from non-16 hrHPV+ women (n = 26, 61.9% versus n = 38, 47.5%; P = 0.181). Similarly, a colposcopist-guided biopsy was obtained more frequently among HPV16+ than among non-16 hrHPV+ women (37, 88.1% versus 58, 72.5%, P = 0.066).
Table 3. Sensitivity and specificity (with 95% CI) of DSI and the colposcopist for the detection of CIN2+ lesions
|Total population||80% (70–88)||77% (67–85)|
|Non-16 hrHPV+||74% (57–87)||67% (50–80)|
|HPV16+||97% (84–100)||100% (69–100)|
|Total population||55% (44–66)||85% (76–91)|
|Non-16 hrHPV+||61% (43–76)||83% (69–93)|
|HPV16+||53% (35–71)||90% (55–100)|
The HG lesion size is given in Table 4, stratified per HPV16 status. High-grade lesion size was defined as the number of pixels indicated as HG by the DSI colour-coded map. Therefore the size of those lesions that were indicated as N/LG by the DSI colour-coded map but as HG by the colposcopist are missing (Table 4). These data show that there were no differences in mean number of HG pixels according to HPV16 status (mean rank 32.86 versus 27.42, P = 0.225 Mann–Whitney U test). However, CIN2+ lesions defined as HG by the colposcopist were significantly larger than those that were defined as N/LG by the colposcopist (mean rank 35.74 versus 21.02, P = 0.001 Mann–Whitney U test).
Table 4. Number of HG pixels of missed (N/LG) and detected (HG) HPV16+ and non-16 hrHPV+ CIN2+ lesions
|Non-16 hrHPV+CIN2+(n = 38)|
|Median HG pixel number [range]|| n = 10||n = 28 |
2632 [40–65 796]
|n = 15 |
|n = 23 |
4350 [20–65 796]**
|HPV16+CIN2+(n = 32)|
|Median HG pixel number [range] ||n = 1||n = 31 |
680 [4–24 948]
|n = 15 |
562 [4–10 225]***
|n = 17 |
1636 [8–24 948]
Our data are in partial agreement with previous findings that visual appearance among HPV16+ women is more defined, although this was previously thought to be irrespective of the severity of the lesion.10 In this study we demonstrate that this effect is dependent on final histology. Furthermore, we add to these data by using a clinically significant outcome measure of N/LG versus HG lesions, instead of an impression of normal versus low-grade or worse (LG+).10
The facts that colposcopic impression was recorded before biopsy taking, and that random biopsies were collected from all women in addition to colposcopically guided biopsies, add credit to our results. Histology in our study is a more reliable ‘gold standard’ than in a previous study in which only colposcopically guided biopsies were taken.10 Furthermore, our cohort of women were referred for colposcopy because of abnormal cytology or follow up of CIN lesions and this reflects the actual population referred for colposcopy in regular clinical practice. We find a relatively high percentage of HPV-negative women with a HG lesion (n = 10, 22.8%) compared with other studies. This might have been caused by a false-negative HPV test because two of these ten women, with a CIN3 lesion, did indeed have a positive hrHPV test on the biopsy sample. An additional explanation could be that the scrapes used for HPV testing were taken directly before colposcopy, and were therefore taken with more caution, resulting in a false-negative test result. However, to keep the analysis uniform throughout the study, we did not correct the status of these women to hrHPV+.
Remarkably, we were only able to reproduce the findings of Jeronimo et al.10 using DSI digital colposcopy. With visual colposcopy we did not observe a difference in either sensitivity or specificity for HG lesions caused by different HPV types. Within the limitations of this study we cannot fully explain why DSI classifies lesions more accurately among HPV16+ women. We explored possible explanations such as the number of biopsies that had been taken, but as a random biopsy was collected in all women this could not have influenced the outcome. Furthermore, a possible confounder, cytology status, has been shown not to differ among the missed and detected lesions among the HPV16+ and non-16 hrHPV+ women. Also the amount of multiple hrHPV infections was similar in both groups, and ‘HG’ pixel count (lesion size) was not related to HPV16 status. Hence a plausible remaining explanation is that HPV16+ lesions show more intense/durable acetowhitening, an effect that is directly used by the DSI software algorithms to map the lesion.18
HPV16 is one of the most oncogenic hrHPV types19–21 so it is reassuring for the colposcopic practice that HPV16-caused lesions can be effectively detected by DSI digital colposcopy. In this era of vaccination of young women against HPV16, the prevalence of HPV16 infections is expected to decrease; however, it is not likely that this will affect the population over the next 15–20 years, when the vaccinated women of today come of age. It has been argued10 that this reduction may eventually lead to the deterioration of colposcopic performance, because this depends largely on the experience of the practitioners. New technologies such as DSI digital colposcopy will maintain colposcopic performance to a high standard. Furthermore, it may be clinically relevant that DSI seems to detect smaller cervical lesions. It can be speculated that by detection and early treatment of these smaller lesions the number of complications of treatment (premature birth and premature rupture of membranes during pregnancy) can be diminished. However, it remains to be elucidated how relevant the detection of small HG lesions is because small HG lesions are more likely to regress spontaneously than large HG lesions.
We studied the agreement between colposcopic impression and final histology for HPV16+ and non-16 hrHPV+ women in a prospective cohort study in which women referred for colposcopy were evaluated by both conventional colposcopy and digital colposcopy using the DSI digital colposcope.8 It has been shown that the DSI digital colposcope improves the sensitivity of colposcopy.7,8 Our main finding in this study is that lesions are better detected among HPV16+ women than among non-16 hrHPV+ women at colposcopy with DSI and that this effect is associated with CIN grade. Furthermore, we show that the HPV16 status does not affect the performance of the colposcopist. It is mainly the lesion size that determines the accuracy of the colposcopic impression, and, as generally anticipated, the colposcopic performance is worse when HG lesions are smaller.
Disclosure of interests
Besides CB and EP, none of the authors received financial support from DySIS Medical Ltd (formerly Forth Photonics Ltd), for the submitted work. CB was a stockholder and EP is an employee of DySIS Medical Ltd (formerly Forth Photonics Ltd) UK at the time of this study; their spouses, partners or children have no financial relationships that may be relevant to the submitted work; and none of the authors have non-financial interests that may be relevant to the submitted work. JL and MK received support from DySIS Medical Ltd (formerly Forth Photonics Ltd), for investigator meetings. Furthermore, JL and AZ received travel support from DySIS Medical Ltd (formerly Forth Photonics Ltd), to visit a conference on colposcopy.
Contribution to authorship
AZ and JAL drafted the manuscript with assistance of JB for statistical analysis. JAL, AZ, MK, WAtH, GCMG and JWMS were responsible for the colposcopies and collection of the data. PJFS supervised the hrHPV testing. FJvK and CJLMM were responsible for revising the histology samples. RHMV was the project leader and designed the clinical study with JAL, MK, WAtH, CB, EP, PJFS and CJLMM. All authors critically reviewed the manuscript.
Details of ethics approval
The ethics boards of the three participating clinics approved the protocol (number 2007/098). Signed informed consent was obtained from all women before any study procedures. The study was registered in the Dutch trial registry (ISRCTN66112760).
HumaVac (VU University Medical Center, Amsterdam, the Netherlands) and DySIS Medical Ltd (formerly Forth Photonics Ltd), were the funding sources for this trial. The VU University Medical Center provided the personnel and facilities to perform the study. DySIS Medical Ltd (formerly Forth Photonics Ltd) provided the DSI digital colposcope and the patient insurance coverage. Their representatives had a role in the study design and they critically appraised the manuscript, but they had no role in data collection or final data analysis. All authors had full access to all data in the study. The corresponding author (AZ) and last author (RHMV) had the final responsibility for the decision to submit for publication.
We thank ThJM Helmerhorst and WGV Quint for their contribution to the study design. We thank DySIS Medical Ltd (formerly Forth Photonics Ltd) for their technical support. Furthermore, we would like to thank all colposcopists, laboratory personnel and women who have participated in this study.