The present study compared two methods for the detection of severe cervical dysplasia in women with atypical squamous cells of underdetermined significance (ASC-US) cytology; digital colposcopy with adjunctive dynamic spectral imaging (DSI) and conventional colposcopy. IMPROVE-COLPO was a two-arm cross-sectional study of US community-based colposcopy. The active (prospective) arm of this study recruited patients examined by digital colposcopy and adjunctive DSI. Preceding consecutive patients that had been examined with conventional methods were used as historical controls in the retrospective arm of the study after being matched in number to those in the prospective arm by a colposcopist. In the present study, the primary measure was the number of women detected with cervical intraepithelial neoplasia (CIN) grade 3 or worse (CIN3+) following punch biopsy. The study included 1,353 retrospective and 1,226 prospective patients eligible for this analysis who were examined by 146 colposcopists in 42 community-based clinics. The patient baseline characteristics were comparable between the two arms. The average number of biopsies taken per patient was higher among the prospective arm patients (including standard and DSI-assisted biopsies) compared with the retrospective arm control patients (1.21 vs. 0.97 respectively). Biopsy detected 31 patients with CIN3+ [2.29%; 95% confidence interval (CI), 1.56-3.24] in the retrospective arm, and 48 patients with CIN3+ (3.92%; 95% CI, 2.90-5.16) in the prospective arm. The difference in the number of patients detected with CIN3+ in the two arms of the study was 1.62% (95% CI, 0.30-3.04; P=0.022), which corresponds to a 70.9% relative increase in the prospective compared with the retrospective arm. Biopsy appeared less efficient in detecting patients with CIN3+ in the retrospective arm compared with the prospective arm. However, there was no statistically significant difference between the retrospective arm and the prospective arm in terms of: i) Biopsies taken (over the entire population) per patient detected with CIN3+ (42.2 in the retrospective arm vs. 30.8 in the prospective arm; P=0.164) and ii) positive predictive value of using biopsies to identify patients with CIN3+ (2.83 vs. 3.92; P=0.118). Adoption of digital colposcopy with DSI increased the number of biopsies collected from ASC-US patients compared with retrospective controls of standard colposcopy and detected a significantly higher number of patients who were CIN3+. The number of additional biopsies taken in the prospective arm compared with the retrospective arm was too small to explain the increased detection of patients with CIN3+ observed in the prospective arm, suggesting that biopsies in the prospective arm were better at identifying CIN3+.
The increase in cervical cancer screening intervals in the US (
Adjunctive Dynamic Spectral Imaging (DSI) mapping is a standardized way to quantify cervical acetowhitening for colposcopic assessment and biopsy selection, which increased the sensitivity of high-grade lesion detection in previous studies (
To the best of our knowledge, the present analysis was the first to present results collected from the recruitment of 7,555 patients in total. The present study evaluated the impact of digital colposcopy combined with DSI on a well-defined sub-population, specifically the cohort of women recruited in IMPROVE-COLPO with cytology of atypical squamous cells of undetermined significance (ASC-US). The present study may be the largest conducted on this specific cohort, with this prevalent but equivocal cytologic indication representing ~1/3 of women having colposcopy (
IMPROVE-COLPO (NCT 02185599) was an observational, two-arm, cross-sectional study that recruited patients who had been referred for colposcopy across multiple US community-based clinics (
The DSI method, which is integrated as an adjunct into a commercially available digital colposcope (DYSIS; DYSIS Medical Ltd.), standardizes colposcopic imaging, quantifies and maps the acetowhitening to introduce objectivity, and supports assessments and biopsy decisions (
The colposcopists that participated in the present study were those performing colposcopy routinely in each facility, and comprised gynecologic oncologists, obstetrician-gynecologists, nurse practitioners and physician assistants. Colposcopists were all adequately trained for the use of the digital colposcope and interpretation of the DSI map prior to recruiting prospective patients. The inclusion criteria for patients across both arms included age ≥21 years, an abnormal screening result based on guidelines (
The DSI digital colposcope was used for all examinations in patients in the prospective arm. This is a high-resolution digital colposcope offering DSI mapping (
Colposcopists were responsible for all clinical decisions, ensuring that the study would capture the pragmatic, rather than an investigational, use of DSI. Decisions regarding which women required a biopsy, and the number and location of the biopsies, were not protocol-dictated. For each patient, the DSI map was interpreted and followed or ignored for biopsy selection at the colposcopists' discretion.
In order to collect a matched control group (retrospective arm of the study), appointment and billing information was used to identify consecutive patients for each colposcopist. Identified patients were subsequently selected according to study inclusion/exclusions. Data were extracted from the patient charts recorded at each clinic.
Prospective arm recruitment occurred between September 2014 and October 2017. Patients in the retrospective arm (control examinations) were between November 2004 and October 2017, with >96% of them between January 2012 and October 2017 and 80% between January 2014 and October 2017. The data collected comprised basic demographics information, numbers of biopsies taken, indication whether endocervical sampling (ECS) or treatment was performed, and histopathology results. Biopsy results are presented at the single biopsy level, with a few exceptions when multiple biopsy samples had been processed and reported together, in which case they were excluded from any biopsy-level analysis. Biopsies were labeled as ‘standard’, ‘random’ or ‘DYSIS’ (for prospective patients). Histopathology readings, which are the gold standard for analyses, were performed by the laboratories used by the participating facilities, following routine practice.
Although the primary end-point of the IMPROVE-COLPO study was the detection of women with CIN2+, the current study presents an analysis primarily for detection of women with CIN3+, which is a more reliable surrogate for cervical cancer (
Given the observational design of the IMPROVE-COLPO study and the non-investigational colposcopy (in both arms), where no biopsies or excisional treatments (loop electrosurgical excision procedure or conization) outside routine care were requested, the sensitivity of colposcopy and colposcopic biopsy to detect patients with high-grade cervical disease could not be measured because the number of subjects with undetected disease was unknown. For analyses, the ‘disease detection rate’ was used, which was defined as the number of patients with CIN3+ divided by the total number of patients. Rates were first calculated for all the detection methods combined (biopsy/ECS/treatment) as an overview. Subsequently, the detection rates were calculated, analyzed and compared specifically for colposcopic (punch) biopsies. In the retrospective arm, the detection results were also calculated for different time periods. As an indirect measure of specificity of colposcopic biopsy, the proportion of women that underwent biopsy without their biopsies revealing CIN3+ was determined. As indicative measures of biopsy efficiency in each arm, the total number of biopsies taken (over the entire population in each arm) per each patient detected with CIN3+ was calculated, and the positive predictive value (PPV) of biopsy was determined, which corresponds to the proportion of biopsies that identified CIN3+.
All statistical analyses were pre-planned using two-sided 5% Type I error. A two-sided Fisher exact test with a two-sided confidence interval (CI) for the difference was used to compare detection rates. A two-sided Kruskal-Wallis test was used to compare the distribution of the number of biopsies and the number of positive biopsies. An exact binomial test was used to evaluate the incremental gain in the detection rate attributed to DSI within the prospective arm.
IMPROVE-COLPO recruited a total of 7,555 women across its two arms. Among these, 2,587 (34.24%) patients had an ASC-US result and were considered for the present study. These patients were examined by 146 colposcopists in 42 clinics. Data were collected separately for each site and colposcopist, so there was no overlap of recruitment between the two arms at each clinic. No study/study-device related adverse events were reported. Eight patients were excluded for the following reasons: Age <21 years (n=4); history of hysterectomy (n=3); and pregnancy (n=1). A total of 1,353 and 1,226 women were included in the retrospective and prospective arms, respectively (
The clinical characteristics of patients and the distribution of patients according to their detected disease status are presented in
In total, 126 patients with CIN3+ were detected: 56 patients in the retrospective arm (4.14%) and 70 patients in the prospective arm (5.71%) (
Punch biopsy detected 31 women with CIN3+ (55.36% of all women detected with CIN3+) in the retrospective arm vs. 48 women with CIN3+ (68.57% of all women detected with CIN3+) in the prospective arm. ECS was performed on >73% of the patients in both arms (
In the retrospective arm, 911/1,353 women (67.3%) underwent biopsy vs. 862/1,226 women (70.3%) in the prospective arm (P=0.1062; two-sided Fisher exact test;
The average number of biopsies specifically among the sub-group of patients that underwent biopsy was 1.44 (retrospective) vs. 1.71 (prospective) (
Colposcopic biopsy detected 31 women with CIN3+ in the retrospective arm vs. 48 in the prospective arm. The detection rates were 2.29% (95% CI, 1.56-3.24) vs. 3.92% (95% CI, 2.90-5.16) respectively (
The difference in the detection of patients with CIN3+ in the two arms remained significant (P=0.027; two-sided Fisher exact test), also when the additional patients with CIN3+ detected by excision (12 in each arm) after biopsy/ies of CIN2 (and ECS of <CIN2) were taken into account. The detection of patients with CIN3+ was consistently higher in the prospective arm for different age sub-groups (
To indirectly assess the specificity of biopsy, the number of women without CIN3+, specifically among biopsied patients, was analyzed. Of the 911 biopsied women in the retrospective arm, 880 (96.6%) had no CIN3+ biopsy vs. 814 of the 862 biopsied women (94.42%) in the prospective arm, a 2.17% difference (P=0.029; two-sided Fisher exact test; data not shown).
Detection in the retrospective arm was analyzed for the different time periods (data not shown), although patient numbers for the earlier years was too small for a definitive conclusion. For the 50 women (3.7% of the total) examined between 2004 and 2011, before ASCCP screening guidelines (
In a secondary analysis, CIN2+ detection was investigated, and results confirmed a similar trend as with CIN3+, as there were 91 women in the retrospective and 116 in the prospective arm (including 1 detected by random biopsy), which correspond to detection rates of 6.73% (95% CI, 5.45-8.19%) and 9.46% (95% CI, 7.88-11.24%) respectively, a 2.74% difference (95% CI, 0.55-4.93%; P=0.011; two-sided Fisher exact test) and a 40.7% increase in the prospective arm compared with the retrospective arm (data not shown).
Among the 1,226 patients from the prospective arm, the incremental detection of patients with CIN3+ diagnosed by additional DSI-assisted biopsies selected at different locations on the cervix after visual assessment biopsies was analyzed (data not shown). Overall, 659 (53.8%) patients had standard biopsies, and 190 (15.5%) of these patients also had DSI-assisted biopsies. In addition, 196 patients (16%) underwent DSI-assisted biopsies without a standard biopsy, whereas an additional 7 patients (0.6%) underwent random biopsy only and 364 patients (29.7%) had no biopsy taken (data not shown).
Altogether, there were 966 standard biopsies and 482 DSI-assisted biopsies (
The majority of the 17 additional patients with CIN3+ that were detected by DSI-assisted biopsies selected after visual assessment, would have been missed without the use of DSI (data not shown in Tables). For 11 of these patients (64.7%), before observing the DSI map, the colposcopist had confirmed that they would not perform a biopsy; 4 patients (23.5%) had standard visual biopsies that were negative or CIN1 and 2 patients (11.8%) had CIN2 in a standard visual biopsy preceding the DSI-assisted biopsy that detected CIN3+.
A similar trend was observed for the detection of women with CIN2+. Standard visual biopsy detected 75 women, which corresponds to a detection rate of 6.12% (95% CI, 4.84-7.61%). The additional DSI-assisted biopsies detected another 40 women with CIN2+, which increased the detection rate to 9.38% (95% CI, 7.81-11.15%), a 53.3% relative increase over the detection by standard visual biopsy.
In a secondary analysis, the biopsy efficiency to detect CIN3+ was investigated, and the results were compared between the two arms and within the prospective arm (data not shown).
Comparing the total number of biopsies taken in each arm to the number of patients detected with CIN3+, suggests that biopsy was less efficient in the retrospective arm, i.e., more biopsies were required to detect a patient with CIN3+ in the retrospective compared with the prospective arm (42.2 vs. 30.8, respectively); however, this difference was not statistically significant (two-sided P=0.164 for the ratio, computed by inverting the two exact binomial tests).
To compare the PPV of biopsy in the two arms, 1 patient with CIN3 from the retrospective arm was excluded as she had multiple biopsies taken, but all of them had been processed together. CIN3+ was detected in 37/1,306 biopsies in the retrospective arm vs. 58/1,478 in the prospective arm. The PPVs were 2.83% (95% CI, 2.0-3.88%) and 3.92% (95% CI, 2.99-5.04%) in the retrospective and prospective arms, respectively. The 1.09% difference was not statistically significant (95% CI, −0.27-2.46%; P=0.118; two-sided Fisher exact test).
Within the prospective arm, biopsies were considered separately as standard/visual and DSI-assisted, in order to analyze their PPV. Overall, 34/966 standard biopsies and 23/482 DSI-assisted biopsies were CIN3+, with PPVs of 3.52% (95% CI, 2.45-4.88%) and 4.77% (95% CI, 3.05-7.07%), respectively. The 1.25% difference was not statistically significant (95% CI, −0.83-3.75%; P=0.254; two-sided Fisher's exact test).
To compare standard/visual biopsy across the two arms, only the first part of prospective arm examinations (biopsy selection by standard visualization) was considered. The number of standard/visual biopsies per patient was higher in the retrospective arm compared with the prospective arm (0.97 vs. 0.79; P<0.001; two-sided Kruskal-Wallis test); however, the PPV for detecting CIN3+ was comparable to prospective arm biopsies (2.83% vs. 3.52%; P=0.394; two-sided Fisher's exact test), resulting in the comparable detection rates for CIN3+ (2.29% vs. 2.45%; P=0.797; two sided Fisher's exact test).
ASC-US is an equivocal cytologic state with low-risk for CIN3+ (
Biopsies selected based on standard visualization (DSI-assisted biopsies not considered for the prospective arm) had comparable detection rates (2.29 vs. 2.45%), indicating that visual/standard colposcopy may be equivalent in the two arms and that the overall increase in the prospective arm may have been primarily due to the use of adjunctive DSI for biopsy selection. The secondary analyses performed on the biopsy-level did not find significantly different results, which may be due to insufficient numbers, and should therefore be interpreted with caution. However, the results from these secondary analyses suggested that prospective arm biopsies were more efficient than retrospective control arm biopsies. Proportionally more biopsies detected CIN3+ and it took fewer biopsies to identify a patient with CIN3+ in the prospective arm. Furthermore, the DSI-assisted biopsies specifically, exhibited the highest PPV, despite the disadvantage of being selected after the visual assessment, which should have identified the majority of obvious lesions, had been completed.
A previous analysis from the IMPROVE-COLPO study analyzed results from a less specific cohort than in the present study and used the global threshold for high-grade disease (CIN2) (
Comparing populations and findings in the present study of colposcopy in community-based clinics with those of academic institution-based trials (
The present study had some limitations. Firstly, without additional biopsies, excisional treatments or longitudinal follow-up, cases of missed disease cannot be confirmed, precluding the calculation of true sensitivity. However, the relative increase of detection rate in the prospective arm compared with that of the retrospective arm corresponds to an equal increase in sensitivity. Without additional detailed analyses of the histological findings that were out of scope (e.g., adjudication and lesion size comparison in treatment specimen), it is hard to confirm the clinical significance of the present results. However, since the primary analysis in the present study was for CIN3+ and the difference in detection was significant, the results from this study may reflect true impact on patient outcomes. Secondly, a direct comparison between the use of DSI for additional biopsies and the collection of additional ‘standard’ biopsies per patient (
One strength of the present study is that ‘real-world’ data, i.e., data representing routine practice, were collected both in the control arm and in the active arm with pragmatic use of DSI, from consecutive patients and with minimal exclusions. In US community-based colposcopy clinics, in contrast to recommendations (
In conclusion, the results from the present study suggest that digital colposcopy combined with DSI mapping may increase the detection of CIN3+ lesions among women with ASC-US. The results from the present study also suggest that, in order to obtain a detection rate for standard visual colposcopy comparable to that achieved with the use of DSI mapping, colposcopists would likely have to increase the number of biopsies taken per patient by more than the 18.8% observed in the present study. These findings may allow the implementation of changes in clinical practice, and therefore reduce the negative impact on patients and the cost of medical care.
Not applicable.
The present study was funded by DYSIS Medical, Edinburgh, UK. The study sponsor was involved in the study design, the collection, analysis and interpretation of data, the writing of the report and the decision to submit the paper for publication.
All data generated or analyzed during this study are included in this published article.
KEH, MDA and SD were site principal investigators for this study and made substantial contributions towards the acquisition, analysis and interpretation of data. EP and PTL were involved in the design of the study. PTL performed the statistical analyses. EP, PTL and KEH drafted the manuscript. MDA and SD critically revised the manuscript for intellectually important content. All authors read and approved the final manuscript.
The present study did not require a specific Institutional Review Board approval. The IMPROVE-COLPO study was approved by the E&I Review Services, Independence, MO (approval no. 14067), The University of Toledo Biomedical Institutional Review Board, Toledo, OH (approval no. 200864) and the Advocate Health Care Institutional Review Board, Downers Grove, IL (approval no. 6248). All patients in the prospective arm signed informed consent prior to participation, and consent was waived by the Institutional Review Boards for patients in the retrospective arm.
Not applicable.
KEH, PTL, MDA and SD declare that their institutions received grants from DYSIS Medical for the performance of the study. EP is an employee of DYSIS Medical. The study sponsor was involved in the study design, the collection, analysis and interpretation of data, the writing of the report and the decision to submit the paper for publication.
American Society for Colposcopy and Cervical Pathology
National Cancer Institute
dynamic spectral imaging
cervical intraepithelial neoplasia
atypical squamous cells of underdetermined significance
Institutional Review Board
human papillomavirus
endocervical sampling
positive predictive value
confidence interval
low-grade squamous intraepithelial lesion
ASC-US LSIL triage study
Kaiser Permanente Northern California
Colposcopic images from a 31-year-old non-Hispanic white/Caucasian patient after a screening result of atypical squamous cells of undetermined significance and human papillomavirus infection. (A) Image of the cervix prior to application of acetic acid (5%). (B) Image of the cervix 30 sec after acetic acid application. (C) Image of the cervix 2 min after acetic acid application. (D) Image of the cervix with the associated DSI map overlay. Extensive acetowhite lesions can be seen in (B and C), with acetowhitening of varying intensity and persistence, as well as mosaicism. Prior to observing the DSI map (D), colposcopic impression was low-grade changes and a single biopsy was selected at 12 o'clock (gray circle with number ‘1’). However, that area, as well as another area at 7 o'clock, were highlighted in the DSI map and suggested high-grade acetowhitening changes (red-yellow-white). Despite guidance for multiple biopsies in such cases, no additional biopsies were taken from this patient. The result from the biopsy was CIN3+ and the patient had subsequent loop electrosurgical excision procedure which was also CIN3. DSI, dynamic spectral imaging; CIN, cervical intraepithelial neoplasia.
Chart describing the patient flow for the retrospective and prospective arms of the present study. The chart presents the inclusions and exclusions numbers, the number of patients that underwent biopsy and the number of patients that were diagnosed with CIN3+ following biopsy. CIN, cervical intraepithelial neoplasia.
Baseline characteristics of the study population.
Characteristic | Retrospective control arm | Prospective arm |
---|---|---|
Patients, n | 1,353 | 1,226 |
Age, years | ||
Median | 34.0 | 34.0 |
Average | 36.5 | 36.7 |
Pre-menopausal |
1,167 (86.3) | 1,074 (87.6) |
Post-menopausal, n (%) | 185 (13.7) | 185 (12.4) |
Insurance status |
||
Private | 1,212 (89.6) | 1,090 (88.9) |
Medicare | 26 (1.9) | 24 (2.0) |
Medicaid/Other | 95 (7.0) | 83 (6.8) |
Uninsured | 18 (1.3) | 25 (2.0) |
Ethnicity |
||
Caucasian | 822 (60.8) | 737 (60.1) |
African-American | 311 (23.0) | 269 (21.9) |
Asian | 33 (2.4) | 31 (2.5) |
Hispanic | 141 (10.4) | 161 (13.1) |
Other | 46 (3.4) | 28 (2.3) |
Referral pathway |
||
ASC-US/HPV+ | 1,266 (93.6) | 1,137 (92.7) |
Persistent ASC-US | 87 (6.4) | 89 (7.3) |
Cases with missing data not included. ASC-US, atypical squamous cells of undermined significance; HPV, human papillomavirus.
Detection rates of patients with CIN3+ identified by directed biopsy in the two study arms overall and by age group. Within the prospective arm, data are also shown separately for standard and incremental DSI-assisted biopsies.
Retrospective control | Prospective arm | ||||||||
---|---|---|---|---|---|---|---|---|---|
Age, years | Standard Biopsy | Total | Standard biopsy | DSI-assisted biopsy | Difference |
P-value | Relative gain, % | ||
Total, n | 1,353 | 1,226 | |||||||
Detection rate, % | 2.29 | 3.92 |
2.45 | 1.39 | 1.62 | 0.022 | 70.9 | ||
21-24 | |||||||||
Patients, n | 165 | 118 | |||||||
Detection rate, % | 1.21 | 5.08 | 3.39 | 1.69 | 3.87 | 0.071 | 319.5 | ||
25-29 | |||||||||
Patients, n | 299 | 261 | |||||||
Detection rate, % | 3.01 | 4.21 | 3.07 | 1.15 | 1.20 | 0.498 | 40.0 | ||
>29 | |||||||||
Patients, n | 889 | 847 | |||||||
Detection rate, % | 2.25 | 3.66 |
2.13 | 1.42 | 1.41 | 0.089 | 62.7 |
Differences are between detection rates in the retrospective control arm and the total of the prospective arm with P-values calculated using a two-sided Fisher's exact test
Total includes a 36-year-old CIN3+ patient identified by Random biopsy in the prospective arm. CIN, cervical intraepithelial neoplasia; DSI, Dynamic Spectral Imaging.