1. A prospective U.S. ovarian cancer screening study using the risk of ovarian cancer algorithm (ROCA) K. H. Lu, S. J. Skates, T. B. Bevers, W. Newland, R. G. Moore, L. Leeds, S. Harris, O. W. Adeyinka, H. A. Fritsche, R. C. Bast

2. Ovarian cancer is the most lethal gynecologic cancer Greater than 75% of cases present with advanced stage disease, when cure rates are < 30% If caught at Stage 1 or 2, cure rates are 60-90% No effective screening methods have been validated Ovarian Cancer

3. Prevalence of ovarian cancer in post-menopausal women is 1 in 2,500 Minimal requirements for ovarian cancer screening in the general population To achieve a positive predictive value > 10% High sensitivity > 75% VERY high specificity > 99.6% Ovarian Cancer

4. CA-125 Limited sensitivity False positives Transvaginal sonography (TVS) Improved sensitivity Increased false positive rate Increased expense 2 stage strategy: Risk of Ovarian Cancer Algorithm (ROCA) incorporates age and CA-125. Risk re-calculated with each new CA-125 test Those with high scores referred to TVS Limitations of CA-125 or TVS alone

5. Annual CA 125 ROCA algorithm Two stage screening Courtesy of U. Menon and I. Jacobs

6. Annual CA 125 ROCA algorithm NORMAL or “LOW” Two stage screening Courtesy of U. Menon and I. Jacobs

7. Annual CA 125 ROCA algorithm SURGERY ABNORMAL or “HIGH” NORMAL or “LOW” Two stage screening Courtesy of U. Menon and I. Jacobs TVS

8. Annual CA 125 ROCA algorithm SURGERY ABNORMAL or “HIGH” “INTERMED” NORMAL or “LOW” Repeat CA 125 Two stage screening Courtesy of U. Menon and I. Jacobs TVS

9. The purpose of this study was to assess the specificity and positive predictive value (PPV) of a 2- stage screening strategy to detect ovarian cancer in post-menopausal women : Tailored to each individual woman’s baseline Only a small fraction of women referred to transvaginal sonography (TVS) and clinical evaluation Purpose

10. Prospective, single arm Initiated in 2001 Multicenter 1. M. D. Anderson, Houston TX 2. Women’s Hospital, Houston TX 3. University of Texas Family Practice, Houston TX 4. Women and Infants Hospital, Providence RI 5. Baylor Sammons Breast Center, Dallas TX 6. John Stoddard Cancer Center, Des Moines, IA 7. Geffen Cancer Center, Vero Beach, FL Data center for study – Massachusetts General Hospital Methods

11. Eligibility criteria Post-menopausal women ages 50-74 Have at least 1 ovary No significant family history of breast or ovarian cancer Cancer-free and no treatment in past 12 mos, aside from hormonal adjuvant therapy Methods

12. Women underwent annual CA-125 blood tests CA-125 II assayed centrally using ROCHE platform CA-125 values were analyzed with ROCA algorithm ROCA recommendation communicated to primary physician and patient TVS performed at study centers Based upon TVS and ROCA, decision for surgery made by physician at study center and patient Methods

13. Results Total number of participants = 3,252 women Total number of screen years = 10,679 years Average number of screen years per woman = 3.3 years Median age 59, with a range of 50-74

14. Characteristic n% Ethnicity Caucasian Black Hispanic Asian Other 2750 264 139 64 35 84.6 8.1 4.3 2.0 1.1 Ever pregnant259486.8 Ever use OCP238879.1 Ever use HRT61620.6 Hx breast cancer42816.6 Baseline characteristics

15. Results Average annual rates: Normal risk “return in 1 year” = 92.6% Intermediate risk “repeat CA-125 in 3 mos = 6.5% High risk “TVS + referral” = 0.9% Overall: Normal risk “return in 1 year”, n= 2666 (82%) Intermediate risk “repeat CA-125 in 3 mos, n= 501 (15.4%) High risk “TVS + referral”, n=85 (2.6%)

16. Study-directed surgeries 8 surgeries  5 early stage cancers 3 high grade invasive Stage IC Stage IIB 2 borderline (LMP) Stage IA  3 without ovarian cancer 2 benign ovarian tumors 1 with no ovarian abnormality (2 months later found to have early stage endometrial cancer)

17. Study-directed surgeries Study Directed Surgeries (n=8) Pt Age at entry 1 st CA125 # annual CA125 CA 125 resulting in “High Risk” TVS resultsSymptomsFindings at surgery 16829174 abnormal No Stage 1 serous LMP 2*649615 abnormal No Stage 1 serous LMP 3*6312224 abnormal No Benign (cystadenoma) 4*5512322 abnormal No Stage IIB high grade invasive cancer 5*53133 18 170 #1: normal #2: abnormal GI Stage IC high grade invasive cancer 669750 abnormal No Benign (cystadenoma) 765193340 abnormal No Stage IC high grade invasive cancer 86045059 abnormal No 1 st surgery: no ovarian disease (sigmoid nodule with necrotic component); 2 nd surgery: endometrial cancer * Detected by ROCA only, and not by CA-125 >35

18. Stage IIB high grade invasive cancer Stage IC high grade invasive cancer Examples of CA 125 trends Patient classified as “Low Risk”

19. Specificity and Positive Predictive Value Specificity 99.9%, 95%CI [99.7%, 99.98%] Positive predictive value 37.5%, 95% CI [11.1%, 100%] - No more than 3 operations to detect 1 case of invasive ovarian cancer Study was not powered to determine sensitivity - 2 borderline cases were not detected by ROCA algorithm - 0 invasive ovarian cases were missed

20. Discussion United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) –randomized 200,000 post-menopausal women, with 50,000 undergoing 2-stage ROCA algorithm screening –prevalence data from 1 st screen published in Lancet Oncology 2009 UK (prevalence data) and US ovarian cancer screening trials demonstrate similar proportions of women triaged: –“Normal risk” annual CA125 (90.9% vs 92.6%) –“Intermediate risk”, repeat CA125 in 3 mos (8.6% vs 6.5%) –“High risk”, TVS and referral (0.5% vs 0.9%)

21. Discussion UK (prevalence data) and US ovarian cancer screening trials: Specificity (99.8% vs 99.9%) PPV for invasive cancer (35.1% vs 37.5%) Stage distribution in UK (prevalent cases): 16 of 34 (47%) invasive cancers were Stage 1, 2 Stage distribution in US (incident cases): 3/3 (100%) invasive cancers were Stage 1, 2

22. Conclusions In this 9-year screening trial of 3,252 women over age 50, the ROCA algorithm followed by TVS demonstrated very few false positives (specificity = 99.9%) Fewer than 1% of women required a TVS each year Overall, the positive predictive value was 37.5%, which greatly exceeded the goal of 10%; i.e. no more than 3 operations will be required to detect 1 invasive ovarian cancer 5 ovarian cancers were detected, all early stage (3 invasive, high grade and 2 borderline)

23. Conclusions 2-stage strategy for ovarian cancer screening is feasible in a U.S. population. Not practice-changing at this time - await the results of the definitive trial that examines mortality as an endpoint, which is currently ongoing in the United Kingdom Future directions -4 marker panel (CA-125, HE4, CEA, VCAM) -“Point of contact” assay

24. Trial Participants Mary Hernandez, RN Deepak Bedi, MD Cynthia Deverson Connie Teodoro Nora Horick, PhD Charlotte Sun, PhD Pati Berger, RN Cindy Brizard Veronica Solano Cristina Vaida, RN Funded by the M. D. Anderson Ovarian Cancer Specialized Program of Research Excellence P50 – CA083639 Golfers Against Cancer Tracey Jo Wilson Foundation Mossy Foundation Shader Family Foundation Norton Fund Acknowledgements