1 / 20

Evaluation of Two-Tiered LAST Classification in Cervical Lesions Reporting

Investigate impact and underreporting caused by LAST terminology. Audit pathology reports and assess criteria with NLP and AI.

christineh
Download Presentation

Evaluation of Two-Tiered LAST Classification in Cervical Lesions Reporting

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Evaluating the Utilization of Lower Anogenital Squamous Terminology (LAST) Two-tiered Classification for High-Grade Preinvasive Cervical Lesions and Its Impact on Reporting Meichin Hsieh, PhD, MSPH, CTR Louisiana Tumor Registry/Epidemiology Program, School of Public Health, Louisiana State University Health Sciences Center, New Orleans, LA NAACCR/IACR 2019 Conference, Vancouver, BC, Canada June 13, 2019

  2. Background • Abstracting in situ of cervical cancer was no longer required by the standard setters in the United States since 1996. • Worldwide, over 95% of cervical neoplasia are HPV-related • In 2006, HPV vaccination was approved by FDA for females aged 9-26 first. • In 2011, CDC funded 4 state registries to participate in the Pre-invasive Cervical Cancer Surveillance (PCCS) Pilot Project • Assess the association of HPV vaccine usage with precancerous cervical lesion incidence in statewide populations • Collect pre-invasive cervical cancer including CIN3 (3-tiered classification system), CIS, AIS, and severe dysplasia diagnosed in 2009 and 2010 • High-grade neoplasm/lesion NOS was not reportable term • CDC continue to fund 4 state registries to collect cervical in situ cancer as routine since then

  3. Background cont’d • In 2012, LAST Standardization Project recommended the 2-tiered classification system, low-grade and high-grade squamous intraepithelial lesions (SIL), for documenting HPV-associated non-invasive cervical lesions in the pathology reports and using p16 IHC staining to verify high-grade as needed. • Cervical lesions classified exclusively as high-grade SIL using this system were not initially reportable as pre-invasive cervical cancer in the PCCS project. • In 2018, Louisiana Tumor Registry (LTR) conducted an audit to evaluate the use of this classification and p16 IHC test in pathology labs and reports.

  4. Objectives • Investigate the use of the 2-tiered terminology and p16 IHC testing on cervical biopsy specimens by laboratories; • Assess the search criteria needed to identify advanced pre-invasive cervical lesions; and • Assess the impact of underreporting cervical lesions caused by terminology changes.

  5. Methods • Survey on pathology labs • Selected 10 labs with a high volume of precancerous cervical cases in Louisiana • Survey was conducted via a phone interview • Survey questions

  6. Audit Software and Sampling Pathology Reports • AIM E-Path Reporter and audit software • Natural language processing (NLP) to interpret the content of pathology reports based on the provided terminologies (search terms) • Artificial intelligence (AI) engines perform content coding and report selection • 5 Laboratories that used both the 2-tiered and 3-tiered term and also used the AIM E-Path reporting system were chosen for the QC audit • Pathology reports received in 2015 with a cervical specimen from a biopsy were included in the sample selection • Randomly selected equal number of positive and negative pathology reports with maximum up to 500 reports from each lab

  7. Screening Pathology Reports • Eligibility and search criteria Pre-2019 Criteria 2019 Criteria • Search Criteria • ICD-O-3: C53.x • Cervical biopsy specimen • CIN3, CIS, AIS, grade 3, any in situ epithelial tumors, and/or severe dysplasia documented • Search Criteria • Pre-2019 criteria • high-grade, high-grade squamous intraepithelial lesions (HSIL or HGSIL), CIN2, CIN2-3 (or CIN2/3), p16 IHC test with cervix • Reportable Terms • CIN3 • CIS • AIS • Severe dysplasia • Reportable Terms • Pre-2019 terms • HSIL/HGSIL • High-grade • CIN2-3 with p16+ • CIN2 with p16+

  8. Manual Review Processing • Positive and negative pathology reports flagged by AIM software were reviewed by a clinician and/or a certified tumor registrar (CTR) • Recoded positive (reportable) terms based on 2019 criteria for each reportable report • Collected and recorded the presence of p16 IHC testing and the subsequent results of p16 IHC staining. • Categorized into three terminology subgroups: • Terms based on pre-2019 reportable terms only (CIN3, CIS, AIS, and severe dysplasia); • Terms based on new reportable terms only (HSIL, High-grade, and CIN2 or CIN2-3 with positive p16 IHC staining); and • Combination.

  9. Data Analysis • Proportion of positive terminologies for reportable precancerous cervical cases and percentages of pathology reports with the p16 test performed by pathology laboratory were generated. • Kappa statistic, positive predictive value (PPV) and negative predictive value (NPV) based on AIM auto-coding versus manual coding (as the gold standard) were computed to assess the agreement, predictability and degree of discrepancy for reportability. • Used chi-square test to assess the association between p16 IHC testing and terminology group.

  10. Flowchart of audit processing Phone Interview 9 labs Select 5 labs for audit Use pre-2019 search criteria in AIM filter Randomly sample: 987 positive & 987 negative Add new search criteria in AIM filter AIM auto-coded: Positive: 1,273; Negative: 701 Manual review End Non-reportable Reportable Combination New terms Pre-2019 terms

  11. Results: Survey

  12. Agreement between AIM auto-coded and manual-coded • Based on 2019 new search criteria (pre-2019 plus high-grade term) • AIM auto-coded: 1,273 reportable & 701 non-reportable cases • Manual review: 822 reportable & 1,152 non-reportable • 822 reportable cases, 475 (57.8%) identified by pre-2019 terms and 347 (42.2%) solely by new terms • Percentage of agreement is 77.2% with Kappa statistic of 0.5642 (moderate agreement). • PPV 0.65 (95% CI: 0.62-0.67) • NPV 1.00 ((5% CI: 0.99-1.00)

  13. Frequency distribution of positive (reportable) terminology by pathology laboratory (N=822) Pre-2019 2019 terms New terms

  14. Percentage of reportable term by subgroup and pathology laboratories in Louisiana (n=822) • For all labs combined, 84% of reportable pathology reports contained new terminology • Half (42.2%) were solely based on new terms • Rang of reportable terms use • Pre-2019 terms only: 3.9% - 49.1% • New terms only: 19.3% - 47.3% • Combination: 27.7%-54.9%. • Labs 4&5 were less likely to use pre-2019 terms only. *Combination (2019 reportable terms) indicates that both pre-2019 and new terms documented in the pathology report.

  15. Proportion of p16 IHC testing by terminology group P<0.0001

  16. Summary of Results • After adding the HSIL (HGSIL)/ high-grade term in search criteria, positive reports increased by 29% for manual review. • 84% of reportable pathology reports received in 2015 contained the 2-tiered terms. • Percentage of cervical biopsy specimens that had a p16 IHC test performed was 13.6% • Almost all new cases based on new terminology can be identified either by HSIL/HGSIL or high-grade term. • By including new terminologies, we were able to add 347 new cases with 73% increase in reportable advanced precancerous cervical lesions.

  17. Conclusions • The Audits findings helped to • Define the new eligibility criteria for reportable precancerous cervical cases • Highlight that both the 2-tiered and 3-tiered nomenclature are needed to ensure complete identification of all advanced precancerous cervical cases • Although the term HSIL/high-grade lesion include CIN2 (CIN2-3) or moderate dysplasia, its association with HPV and likelihood to predict cervical cancer progression makes it important to collect. • In order to align with the current practice and be able to compare data collected before 2019 and after, the 2019 reportable terms include • Pre-2019 reportable terms (AIS, CIN3, CIS, severe dysplasia) • HSIL/high-grade • CIN2 or CIN2-3 with positive p16 IHC tests

  18. Co-authors • LTR • Xiao-Cheng Wu, MD, MPH • Christina Lefante, MPH • Mary Anne Lynch, MPH • Natalie Gomez, BSN, RN • Brent Mumphrey, BS • Lauren Maniscalco, MPH • Rachna Jetly-Shridhar, MD • CDC • Elizabeth Van Dyne, MD, MPH • Jean A. Shapiro, PhD • Paran Pordell, MPH • Mona Saraiya, MD, MPH

  19. Acknowledgments • Louisiana hospital reference and free-standing pathology laboratories and pathologists • Funding source: • Centers for Disease Control and Prevention under National Program of Cancer Registries Component #2 cooperative agreement number U58DP006332.

  20. Contact Information: mhsieh@lsuhsc.edu

More Related