Prospective Study Using the Risk of Ovarian Cancer Algorithm to Screen for Ovarian Cancer
http://www.100md.com
《临床肿瘤学》
the Department of Gynecological Oncology, Institute of Women's Health, University College London, London
Biostatistics Centre, MA General Hospital, and Harvard Medical School, Boston, MA
US Department of Gynecological Oncology, Barts and The London NHS Trust, London Department of Clinical Biochemistry, University College London Hospital, London Department of Experimental Therapeutics, MD Anderson Cancer Center, Houston, TX
ABSTRACT
PURPOSE: To evaluate prevalence screening in the first prospective trial of a new ovarian cancer screening (OCS) strategy (risk of ovarian cancer or ROC algorithm) on the basis of age and CA125 profile.
PATIENTS AND METHODS: Postmenopausal women, 50 years were randomly assigned to a control group or screen group. Screening involved serum CA125, interpreted using the ROC algorithm. Participants with normal results returned to annual screening; those with intermediate results had repeat CA125 testing; and those with elevated values underwent transvaginal ultrasound (TVS). Women with abnormal or persistently equivocal TVS were referred for a gynecologic opinion.
RESULTS: Thirteen thousand five hundred eighty-two women were recruited. Of 6,682 women randomly assigned to screening, 6,532 women underwent the first screen. After the initial CA125, 5,213 women were classified as normal risk, 91 women elevated, and 1,228 women intermediate. On repeat CA125 testing of the latter, a further 53 women were classified as elevated risk. All 144 women with elevated risk had TVS. Sixteen women underwent surgery. Eleven women had benign pathology; one woman had ovarian recurrence of breast cancer; one woman had borderline; and three women had primary invasive epithelial ovarian cancer (EOC). The specificity and positive predictive value (PPV) for primary invasive EOC were 99.8% (95% CI, 99.7 to 99.9) and 19% (95% CI, 4.1 to 45.6), respectively.
CONCLUSION: An OCS strategy using the ROC algorithm is feasible and can achieve high specificity and PPV in postmenopausal women. It is being used in the United Kingdom Collaborative Trial of Ovarian Cancer Screening and in the United States in both the Cancer Genetics Network and the Gynecology Oncology Group trials of high-risk women.
INTRODUCTION
Ovarian cancer is the fourth most common cause of death from cancer in women in the developed world. It is mainly a disease of women over 50 and is characterized by vague, nonspecific symptoms, presentation at an advanced stage, and poor overall survival. Despite advances in treatment, there has been little change in the mortality rate (http://seer.cancer.gov/faststats/html/inc_ovary.html). Concerted efforts are, therefore, being made to develop an effective strategy for early detection.
A large randomized controlled trial (RCT) of ovarian cancer screening (OCS) is underway in the United Kingdom, involving 200,000 postmenopausal women (UK Collaborative Trial of Ovarian Cancer Screening, UKCTOCS, www.ukctocs.org). This trial incorporates a novel screening strategy utilizing CA125 and ultrasound developed during the last decade. In 1993, we demonstrated that a multimodal screening strategy to detect ovarian cancer, using serum CA125 as the first line test and pelvic ultrasound as the secondary test can achieve high specificity (99.9%) and positive predictive value (PPV; 26.8%).1,2 In the RCT that followed, median survival was significantly increased in women with ovarian cancer in the screened group (72.9 months) when compared with the control group (41.8 months).3 In both trials, the serum CA125 was interpreted using a fixed cutoff of 30 U/mL. Analysis of the data revealed that serial CA125 values before detection of ovarian cancer exhibited increasing levels following a change-point while for almost all other subjects CA125 values had a flat profile that fluctuated around an individual's own baseline level. Interpreting this additional information in serial CA125 values using longitudinal statistical models retrospectively increased the sensitivity for detection of ovarian cancer from 70% to 86%,4 while maintaining a high level of first line specificity (98%), exceeding that of the fixed CA125 cutoff for referral to ultrasound. The model estimates an individual's risk of ovarian cancer (ROC), starting with an estimate on the basis of age and modified by the relative fit of the serial CA125 profile to the change-point model estimated from known cases compared with the flat profile model estimated from known controls.5 In women without ovarian cancer, the expected CA125 profile is flat at an individual's baseline level, while in women with undiagnosed ovarian cancer, the expected CA125 profile is initially flat at the individual's baseline level, but after the change-point increases significantly, presumably because of growth of the tumor. The risk estimate is presented as the risk of the individual having a diagnosis of ovarian cancer were the individual to be operated on immediately such that 2% represents a one in 50 chance. The algorithm recommends screening actions on the basis of the calculated risk, which is updated after each new CA125 measurement for an individual. The algorithm is implemented as follows. Women with a significantly increasing CA125 value above their own baseline will have an elevated risk and be referred to ultrasound, while women with a relatively flat CA125 profile will have a normal risk and be referred to an annual CA125 test, and women with an intermediate CA125 profile will have an intermediate risk and be referred to a repeat CA125 test within a few months. Parallel to this effort, analysis of the ultrasound data from the trials established that refining the ultrasound criteria could increase specificity.6,7 Complex ovarian morphology was the best predictor of ovarian cancer in asymptomatic postmenopausal women found to have an elevated serum CA125 level. This was incorporated into the new strategy.
A prospective screening trial using the ROC algorithm was then undertaken to inform the design of a definitive RCT on the scale of UKCTOCS. This report describes the outcome of the prevalence screen in this trial and provides the first data on prospective use of the ROC algorithm in OCS.
PATIENTS AND METHODS
Patients
Approval was obtained from the local ethics committee. Women from England, Scotland, and Wales were invited to volunteer via articles in the press and leaflets distributed through collaborating general practices and occupational health departments of major companies. In addition, invitations were sent to those who had participated in our previous screening studies.1,3 Women who expressed interest were sent a detailed fact sheet describing the study and a recruitment appointment at the Ovarian Cancer Screening Unit at Bart's Hospital or at their general practice. At these appointments, the study design was explained and volunteers completed a questionnaire and gave written consent. Volunteers were eligible if they fulfilled the following criteria: (a) Age 50 years and (b) more than 12 months amenorrhea, following a natural or surgical menopause or more than 12 months of hormone replacement therapy commenced for menopausal symptoms. The exclusion criteria were: (a) history of bilateral oophorectomy, (b) active malignancy (women with a past history of malignancy were eligible if they had no documented persistent or recurrent disease), (c) increased risk of ovarian cancer because of familial predisposition—exclusion criteria were entry criteria for the UK Familial Ovarian Cancer Screening Study, and (d) previous history of ovarian cancer.
Trial Design
Eligible patients were randomly assigned with equal chance to either the control group or screening group using a computerized random number generator. All patients randomly assigned to the screening group underwent primary screening (level I) with CA125. If an abnormality was detected the women had secondary screening (level II) using CA125 and transvaginal ultrasound (TVS) of the ovaries. A customized database was built for the trial that assessed eligibility and undertook randomization, ROC calculation, classification of results, and letter printing.
Screening Tests
The women in the screening group were posted venipuncture equipment and asked to visit their general practitioners for venipuncture. Blood samples were sent to the Ovarian Cancer Screening Unit by first class mail, centrifuged at 4,000 rpm for 10 minutes and the serum separated, aliquoted, and stored at –20°C. All blood samples received more than 56 hours after venipuncture were discarded and repeat samples requested. Serum CA125 levels were determined by commercial radioimmunoassay (CA 125II kit, Centocor, Malvern, PA) within 2 weeks of receipt of the blood sample. The CA125 values were entered into the trial database, which calculated the ROC (eg, a risk of 2% implied a risk of one in 50 of ovarian cancer), and then generated the appropriate letter.
TVS was performed either at the Ovarian Cancer Screening Unit or at a collaborating center by a consultant radiologist or experienced ultrasonographer. Where TVS was not possible or acceptable, a transabdominal scan was performed. Ovarian morphology and dimensions were assessed and volume determined using the formula for an ovoid (d1 x d2 x d3 x 0.532). Ovarian morphology was classified as normal if the ovary was of uniform hypoechogenicity and smooth outline, simple cyst if a single, thin walled, anechoic cyst with no septa or papillary projections was detected, and complex if there was nonuniform echogenicity because of cystic and solid areas in the ovary, multiple cysts, or ascites. The larger of the two ovaries with the most abnormal morphology was used to classify scans as shown in Table 1.
Screening Strategy
Level I screen (CA125). Women in the screen arm underwent an annual serum CA125 assay interpreted using the ROC calculation. On the basis of the risk value, patients were allocated to one of three groups and managed as detailed below:
Normal risk (< one in 2,000). Patients were informed that their results were normal.
Intermediate risk (one in 2,000 to one in 500). Patients were recalled for a repeat venipuncture. The interval of recall varied between 6 weeks and 6 months and was inversely related to the risk estimate. Management following repeat testing depended on the recalculated risk value, which incorporated the latest CA125 result.
Elevated risk (> one in 500). Patients were recalled for a level II screen.
Level II screen (TVS and CA125). Women underwent a scan of their ovaries and serum CA125 assay. On the basis of the results of these tests, they were managed as presented in Table 2.
Normal scan and risk less than one in 25. Patients were informed that their results were normal.
Normal scan and risk one in 25 to one in five. Serum CA125 was repeated and the risk reassessed. Subsequent management was determined by the same risk criteria as described above for level 1 screens.
Equivocal or unsatisfactory scans irrespective of the risk value or normal scan with risk more than one in five. TVS and CA125 were repeated after ruling out other conditions associated with a CA125 elevation. The patient was referred for surgery if the scan findings were persistently equivocal or became abnormal.
Abnormal scans irrespective of the risk value. The patient was referred to a gynecologic oncologist for assessment and possible surgical investigation.
Confirmation of Diagnosis
Operative notes and histology reports of those who underwent surgery were reviewed to confirm diagnosis and the stage, grade, and histology of any cancer diagnosed. Further information regarding the women diagnosed with ovarian cancers was obtained from the surgeon, medical oncologist, and general practitioner.
Follow-Up
All volunteers who did not undergo surgery or had not withdrawn from the study were contacted a year later for further screening and were sent a postal follow-up questionnaire 3 years after recruitment.
Analysis
The end point of the prevalence screen was defined as surgery, following referral to a specialist or return to annual screening. The prevalence screen, therefore, consisted of a single or series of serum CA125/scans culminating in surgery or return to annual screening. Index cancers were defined as primary invasive epithelial carcinomas of the ovary and fallopian tube. For the purpose of this analysis, all patients have been censored 1 year after the date of the final blood or scan test related to their prevalence screen. Their status, with regard to a diagnosis of an index cancer, was determined on the date of censorship.
Role of the Funding Agencies
The charities that funded the study had no role in study design, collection, analysis, interpretation of data, writing of the report, or decision to publish.
RESULTS
A total of 13,582 women were recruited in the United Kingdom between 1995 and 2000. Fifty women provided incomplete data. Sixty six women were ineligible, the majority (51 women) because they were premenopausal. Of the remaining, 6,682 women were randomly assigned to the screen arm. Table 3 summarizes their baseline characteristics. Figure 1 summarizes the overall course of the study.
Level I Screen
After being randomly assigned to the screen group, 150 patients did not provide a blood sample because they had both ovaries removed (seven women), developed a new cancer or other illness (11 women), withdrew (46 women), or did not respond (86 women). Of the 6,532 women who had a blood test, the risk calculation was normal in 5,213 women, intermediate in 1,228 women, and elevated in 91 women. The 1,228 women with intermediate risk values (ie, one in 2,000 to one in 500) had repeat CA125 testing. Of these women, 1,131 women were returned to annual screening, 44 women withdrew, and 53 women were found to have elevated risk (Fig 1). All 144 women (91 + 53) with an elevated risk (ie, risk > one in 500) were offered a level II screen. If a fixed cutoff of CA125 more than 30 U/mL had been used, 413 women would have required a level II screen.
Level II Screen and Further Management
Six of the 144 women in the elevated risk group were found to have nonovarian malignancies—myeloma, hepatocellular carcinoma, colon, breast, endometrial cancer, and adenocarcinoma of unknown primary (ACUP). In the latter case the woman had a risk estimate of more than one in five with normal ovaries on scan. On computed tomography, she had ascites with a large splenic mass, extensive upper abdominal lymphadenopathy, a 2 cm right external iliac mass, and masses in the upper body of stomach and lower lobe of right lung but no pelvic abnormality. Biopsy of a lymph node confirmed metastatic adenocarcinoma and the diagnosis on multidisciplinary review was of abdominal carcinomatosis of unknown primary origin.
The results of the level II screen in the remaining 138 women were classified as shown in Table 4 with the overall outcome summarized in Figure 2.
Normal scan and risk less than one in 25. All 39 patients in this group returned to annual screening.
Normal scan and risk one in 25 to one in five. All 28 women had a further CA125 and were then returned to annual screening.
Equivocal or unsatisfactory scans irrespective of the risk value or normal scan with risk more than one in five. All 58 women had a repeat CA125 and TVS. Six women underwent surgery as the repeat TVS was abnormal (two women) or persistently equivocal (four women). Fifty-two women returned to annual screening.
Abnormal scans irrespective of the risk value. Thirteen women were assessed clinically and 10 women underwent surgery. Three women (risk values < 1%) were managed conservatively; one had a diagnosis of uterine fibroids, a second had a biloculated small ovarian cyst with no sinister features, and the third had multicystic ovaries with the diameter of the largest cyst less than 2 cm. The latter was followed-up with repeat scans and bloods for 6 months during which time the ovarian size decreased.
Surgery in Screen Positives
Overall, 16 women underwent surgery as a result of prevalence screening. Eleven women had surgery at Bart's Hospital, London and five women had surgery in other hospitals after referral by their general practitioner. Seven women had a laparotomy, four women a diagnostic laparoscopy, and five women a laparoscopic salpingo-oophorectomy. Two of the latter women had completion laparotomy as they were found to have ovarian cancer. Postoperatively, one woman (without ovarian cancer) developed bowel obstruction and had to undergo a further laparotomy, with drainage of a pelvic abscess and resection of small bowel, caecum, and ascending colon.
Final Diagnosis in Screen Positives
Ovarian malignancy was detected in five women. One had ovarian recurrence of a breast cancer 16 years after initial diagnosis, one had a borderline papillary serous ovarian carcinoma, and three had primary invasive epithelial ovarian cancers (EOCs) (Table 5). The details of the benign findings in the remaining 11 women are summarized in Table 6.
Follow-Up
Follow-up data at 1 year after the final test related to the prevalence screen was available in 97.5% of the screened women (6,369 women). No additional cases of ovarian cancer were documented during follow-up.
Screening Performance
The screening strategy achieved a PPV of 19% (Table 7) for detection of primary invasive EOC. Specificity was 99.8%, if all participants are included. If only those individuals who were followed for at least 1 year are used to calculate specificity, then it was 99.7%.
DISCUSSION
This report describes the first prospective use of the ROC algorithm in screening for ovarian cancer. The results demonstrate that OCS, using the algorithm is feasible and can achieve high specificity (99.8%; 95% CI, 99.7% to 99.9%) and PPV (19%; 95% CI, 4.1% to 45.6%). The initial algorithm was developed using stored serum from apparently healthy women in the Stockholm OCS study, six of whom were found to have ovarian cancer.8 Retrospective analysis estimated it achieved a specificity of 99.7% and PPV of 16% for detection of ovarian cancers diagnosed within a year of the assay.5 The algorithm was then applied retrospectively to 33,621 CA125 results from 9,233 women from the Bart's OCS trials. When compared with a fixed cutoff for CA125, the area under the curve significantly improved from 84% to 93% (P = .01).4 The results reported here demonstrate for the first time in a prospective study that this novel screening strategy algorithm can achieve the high specificity and PPV required for OCS. These are essential characteristics as positive screens result in surgical investigation, with its attendant morbidity.
The trial was undertaken to inform the design of a definitive OCS trial with mortality as a primary end point and to assess the feasibility and logistic implications of this approach. On the basis of the initial blood test, 18.8% of women had an intermediate risk value. On repeat testing, 92% with an intermediate value were recalculated to be at normal risk and returned to annual screening. The initial protocol did not specify the number of repeat tests, and 252 women were referred for four or more tests (Fig 1). Of the women recalled for four or more tests, 6% (15 out of 252) withdrew from the study compared with 1.5% (18 out of 1,228) who withdrew after their first intermediate risk result (Fig 1). It was noted that the 15 women who withdrew after multiple tests gave the disruption and stress resulting from repeat testing as their primary reason for withdrawing. For the definitive UKCTOCS trial it was, therefore, decided that a maximum of three CA125 tests would be requested, and that women who remained in the intermediate group would then be offered an ultrasound scan. Assessment of anxiety and worry resulting from repeat testing is one of the core objectives of the psychosocial study that accompanies UKCTOCS.
The overall rate of referral for ultrasound assessment of the ovaries was 2.2% (144 of 6,514) using the ROC algorithm, compared with a rate of 6.3% (413 of 6,514), had a fixed cutoff for CA125 (> 30 U/mL) been used. Ultrasound is an expensive and labor intensive procedure, and the higher rate would have a significant impact on cost and logistics of a definitive trial and, if successful, a national screening program.
In the course of the study, six nonovarian cancers were diagnosed as a result of elevated CA125 levels. These included a case of ACUP. The findings on imaging in this particular individual made it unlikely that the primary site was ovarian. However, the classification of ACUP raises a difficult issue in OCS trials. Ovarian cancer, if diagnosed at an advanced stage, may wrongly be classified as ACUP, resulting in lower numbers of ovarian cancer cases (true-positives). This can cause under reporting of ovarian cancer cases in the control arm of a screening trial. It is important that as much information as possible be gathered about such cases and that the number of ACUP cases in the screen and controlled arms are monitored; the numbers should not be significantly different.
Sixteen operations were undertaken following the level II screen and clinical assessment. Eleven patients did not have ovarian malignancy. The major cause for false-positive surgery was benign ovarian cystadenomas and the presence of pelvic adhesions. In two cases this was related to diverticulitis, which is a known cause for raised CA125 levels. Initial laparoscopic assessment made it possible to avoid oophorectomy in women with normal ovaries but dense pelvic adhesions, where the risk of surgical complications may be increased.
Of the five women found to have ovarian malignancy, one had a stage IA borderline serous micropapillary ovarian cancer. There is controversy among trialists about inclusion of borderline tumors (tumors of low malignant potential) as true-positives in OCS trials. Recently, it has been proposed that the serous borderline tumors should be classified as micropapillary serous carcinoma or atypical proliferative serous tumor. The former have a higher incidence of bilateral ovarian involvement, recurrence with shorter progression-free interval, and invasive implants. However, the overall survival of patients in both groups appear to be the same,9 and both these tumors have an overall good prognosis. We have, therefore, continued our practice of not including borderline ovarian carcinomas as true-positives irrespective of their histology as the aim is to develop a strategy that impacts on ovarian cancer mortality.
Ovarian recurrence of a primary breast cancer was detected in one patient. Women who had previously been diagnosed with a malignancy and were in remission were allowed to join the study. A total of 5.12% (689 women) in the screened arm had a personal history of cancer, mainly breast and bowel. As CA125 can occasionally be elevated in recurrences of such malignancies,10 the detection of occasional asymptomatic recurrences is inherent in the study design. The benefits of such detection are uncertain and women with a previous history of cancer should be counseled about this at recruitment.
The prevalence screen reported here involved 6,532 women. Three women were diagnosed with primary invasive EOC, two women stage I, and one woman stage II. No other primary ovarian cancers were reported in these women during the year following their screen. The numbers are too small to comment on sensitivity or stage shift but are consistent with the data from retrospective evaluation of the ROC algorithm. There is preliminary evidence from our previous RCT that early detection may increase long-term survival.3 Follow-up data from prospective single-arm OCS trials also provide evidence of a possible survival benefit.11,12 It is, however, important to note that the ovarian cancers detected by screening in the latter trials included granulosa cell and borderline ovarian carcinomas that have a good overall prognosis.
Large RCTs are required to assess the impact of screening on ovarian cancer mortality in the general population. In the United States the Prostate, Lung, Colorectal, and Ovarian Cancer study involving 74,000 women includes an OCS arm.13-19 The trial reported here has informed the design of the UKCTOCS, which aims to randomly assign 200,000 postmenopausal women, aged 50 to 74 years, in a 1:1:2 ratio to annual ultrasound screening, CA125 screening using the ROC algorithm, and a control group (www.ukctocs.org.uk). UKCTOCS includes a comprehensive evaluation of cost, morbidity, compliance, and acceptability of screening, in addition to the primary end point of ovarian cancer mortality. More than 95,000 women have now been randomly assigned into this study, which will be completed in 2011 and is designed to provide definitive data about the impact of population screening for ovarian cancer.
The algorithm is also being used for screening high-risk women in the US Cancer Genetics Network trial, and in the screening arm of the Gynecology Oncology Group study of women at high risk in the United States. In the UK Familial Ovarian Cancer Screening Study, efforts are underway to revise the current screening strategy to incorporate the ROC algorithm. If screening were shown to be effective in these various trials, it is highly likely that the ROC algorithm that will be part of future OCS programs across the world.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank all the women who participated in the trial. Also our thanks to all the general practitioners whose collaboration was essential to the study.
NOTES
Supported by grants from the UK charities Research into Ovarian Cancer and the Gynecology Cancer Research Fund.
Presented in part at the Annual Meeting of the European Society of Gynecological Oncology, Brussels, Belgium, May 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Jacobs I, Davies AP, Bridges J, et al: Prevalence screening for ovarian cancer in postmenopausal women by CA 125 measurement and ultrasonography. BMJ 306:1030-1034, 1993
Jacobs I, Stabile I, Bridges J, et al: Multimodal approach to screening for ovarian cancer. Lancet 1:268-271, 1988
Jacobs IJ, Skates SJ, MacDonald N, et al: Screening for ovarian cancer: A pilot randomized controlled trial. Lancet 353:1207-1210, 1999
Skates SJ, Menon U, MacDonald N, et al: Calculation of the risk of ovarian cancer from serial CA-125 values for preclinical detection in postmenopausal women. J Clin Oncol 21:206-210, 2003
Skates SJ, Pauler DK, Jacobs IJ: Screening based on the risk of cancer calculation from Bayesian hierarchical change point and mixture models of longitudinal markers. J Am Stat Assoc 96:429-435, 2001
Menon U, Talaat A, Jeyarajah AR, et al: Ultrasound assessment of ovarian cancer risk in postmenopausal women with CA125 elevation. Br J Cancer 80:1644-1647, 1999
Menon U, Talaat A, Rosenthal AN, et al: Performance of ultrasound as a second line test to serum CA125 in ovarian cancer screening. BJOG 107:165-169, 2000
Einhorn N, Sjovall K, Knapp RC, et al: Prospective evaluation of serum CA 125 levels for early detection of ovarian cancer. Obstet Gynecol 80:14-18, 1992
Deavers MT, Gershenson DM, Tortolero-Luna G, et al: Micropapillary and cribriform patterns in ovarian serous tumors of low malignant potential: A study of 99 advanced stage cases. Am J Surg Pathol 26:1129-1141, 2002
Sjovall K, Nilsson B, Einhorn N: The significance of serum CA 125 elevation in malignant and nonmalignant diseases. Gynecol Oncol 85:175-178, 2002
van Nagell JR Jr, DePriest PD, Reedy MB, et al: The efficacy of transvaginal sonographic screening in asymptomatic women at risk for ovarian cancer. Gynecol Oncol 77:350-356, 2000
Sato S, Yokoyama Y, Sakamoto T, et al: Usefulness of mass screening for ovarian carcinoma using transvaginal ultrasonography. Cancer 89:582-588, 2000
Gohagan JK, Prorok PC, Hayes RB, et al: The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: History, organization, and status. Control Clin Trials 21:251S-272S, 2000
Hartge P, Hayes R, Sherman M, et al: Complex ovarian cysts in postmenopausal women are not associated with ovarian cancer risk factors. Preliminary data from the PLCO cancer screening trial. Ann Epidemiol 10:465, 2000
Hasson MA, Fagerstrom RM, Kahane DC, et al: Design and evolution of the data management systems in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:329S-348S, 2000
Hayes RB, Reding D, Kopp W, et al: Etiologic and early marker studies in the prostate, lung, colorectal and ovarian (PLCO) cancer screening trial. Control Clin Trials 21:349S-355S, 2000
Miller AB, Yurgalevitch S, Weissfeld JL: Death review process in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:400S-406S, 2000
O'Brien B, Nichaman L, Browne JE, et al: Coordination and management of a large multicenter screening trial: The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:310S-328S, 2000
Prorok PC, Andriole GL, Bresalier RS, et al: Design of the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:273S-309S, 2000(Usha Menon, Steven J. Ska)
Biostatistics Centre, MA General Hospital, and Harvard Medical School, Boston, MA
US Department of Gynecological Oncology, Barts and The London NHS Trust, London Department of Clinical Biochemistry, University College London Hospital, London Department of Experimental Therapeutics, MD Anderson Cancer Center, Houston, TX
ABSTRACT
PURPOSE: To evaluate prevalence screening in the first prospective trial of a new ovarian cancer screening (OCS) strategy (risk of ovarian cancer or ROC algorithm) on the basis of age and CA125 profile.
PATIENTS AND METHODS: Postmenopausal women, 50 years were randomly assigned to a control group or screen group. Screening involved serum CA125, interpreted using the ROC algorithm. Participants with normal results returned to annual screening; those with intermediate results had repeat CA125 testing; and those with elevated values underwent transvaginal ultrasound (TVS). Women with abnormal or persistently equivocal TVS were referred for a gynecologic opinion.
RESULTS: Thirteen thousand five hundred eighty-two women were recruited. Of 6,682 women randomly assigned to screening, 6,532 women underwent the first screen. After the initial CA125, 5,213 women were classified as normal risk, 91 women elevated, and 1,228 women intermediate. On repeat CA125 testing of the latter, a further 53 women were classified as elevated risk. All 144 women with elevated risk had TVS. Sixteen women underwent surgery. Eleven women had benign pathology; one woman had ovarian recurrence of breast cancer; one woman had borderline; and three women had primary invasive epithelial ovarian cancer (EOC). The specificity and positive predictive value (PPV) for primary invasive EOC were 99.8% (95% CI, 99.7 to 99.9) and 19% (95% CI, 4.1 to 45.6), respectively.
CONCLUSION: An OCS strategy using the ROC algorithm is feasible and can achieve high specificity and PPV in postmenopausal women. It is being used in the United Kingdom Collaborative Trial of Ovarian Cancer Screening and in the United States in both the Cancer Genetics Network and the Gynecology Oncology Group trials of high-risk women.
INTRODUCTION
Ovarian cancer is the fourth most common cause of death from cancer in women in the developed world. It is mainly a disease of women over 50 and is characterized by vague, nonspecific symptoms, presentation at an advanced stage, and poor overall survival. Despite advances in treatment, there has been little change in the mortality rate (http://seer.cancer.gov/faststats/html/inc_ovary.html). Concerted efforts are, therefore, being made to develop an effective strategy for early detection.
A large randomized controlled trial (RCT) of ovarian cancer screening (OCS) is underway in the United Kingdom, involving 200,000 postmenopausal women (UK Collaborative Trial of Ovarian Cancer Screening, UKCTOCS, www.ukctocs.org). This trial incorporates a novel screening strategy utilizing CA125 and ultrasound developed during the last decade. In 1993, we demonstrated that a multimodal screening strategy to detect ovarian cancer, using serum CA125 as the first line test and pelvic ultrasound as the secondary test can achieve high specificity (99.9%) and positive predictive value (PPV; 26.8%).1,2 In the RCT that followed, median survival was significantly increased in women with ovarian cancer in the screened group (72.9 months) when compared with the control group (41.8 months).3 In both trials, the serum CA125 was interpreted using a fixed cutoff of 30 U/mL. Analysis of the data revealed that serial CA125 values before detection of ovarian cancer exhibited increasing levels following a change-point while for almost all other subjects CA125 values had a flat profile that fluctuated around an individual's own baseline level. Interpreting this additional information in serial CA125 values using longitudinal statistical models retrospectively increased the sensitivity for detection of ovarian cancer from 70% to 86%,4 while maintaining a high level of first line specificity (98%), exceeding that of the fixed CA125 cutoff for referral to ultrasound. The model estimates an individual's risk of ovarian cancer (ROC), starting with an estimate on the basis of age and modified by the relative fit of the serial CA125 profile to the change-point model estimated from known cases compared with the flat profile model estimated from known controls.5 In women without ovarian cancer, the expected CA125 profile is flat at an individual's baseline level, while in women with undiagnosed ovarian cancer, the expected CA125 profile is initially flat at the individual's baseline level, but after the change-point increases significantly, presumably because of growth of the tumor. The risk estimate is presented as the risk of the individual having a diagnosis of ovarian cancer were the individual to be operated on immediately such that 2% represents a one in 50 chance. The algorithm recommends screening actions on the basis of the calculated risk, which is updated after each new CA125 measurement for an individual. The algorithm is implemented as follows. Women with a significantly increasing CA125 value above their own baseline will have an elevated risk and be referred to ultrasound, while women with a relatively flat CA125 profile will have a normal risk and be referred to an annual CA125 test, and women with an intermediate CA125 profile will have an intermediate risk and be referred to a repeat CA125 test within a few months. Parallel to this effort, analysis of the ultrasound data from the trials established that refining the ultrasound criteria could increase specificity.6,7 Complex ovarian morphology was the best predictor of ovarian cancer in asymptomatic postmenopausal women found to have an elevated serum CA125 level. This was incorporated into the new strategy.
A prospective screening trial using the ROC algorithm was then undertaken to inform the design of a definitive RCT on the scale of UKCTOCS. This report describes the outcome of the prevalence screen in this trial and provides the first data on prospective use of the ROC algorithm in OCS.
PATIENTS AND METHODS
Patients
Approval was obtained from the local ethics committee. Women from England, Scotland, and Wales were invited to volunteer via articles in the press and leaflets distributed through collaborating general practices and occupational health departments of major companies. In addition, invitations were sent to those who had participated in our previous screening studies.1,3 Women who expressed interest were sent a detailed fact sheet describing the study and a recruitment appointment at the Ovarian Cancer Screening Unit at Bart's Hospital or at their general practice. At these appointments, the study design was explained and volunteers completed a questionnaire and gave written consent. Volunteers were eligible if they fulfilled the following criteria: (a) Age 50 years and (b) more than 12 months amenorrhea, following a natural or surgical menopause or more than 12 months of hormone replacement therapy commenced for menopausal symptoms. The exclusion criteria were: (a) history of bilateral oophorectomy, (b) active malignancy (women with a past history of malignancy were eligible if they had no documented persistent or recurrent disease), (c) increased risk of ovarian cancer because of familial predisposition—exclusion criteria were entry criteria for the UK Familial Ovarian Cancer Screening Study, and (d) previous history of ovarian cancer.
Trial Design
Eligible patients were randomly assigned with equal chance to either the control group or screening group using a computerized random number generator. All patients randomly assigned to the screening group underwent primary screening (level I) with CA125. If an abnormality was detected the women had secondary screening (level II) using CA125 and transvaginal ultrasound (TVS) of the ovaries. A customized database was built for the trial that assessed eligibility and undertook randomization, ROC calculation, classification of results, and letter printing.
Screening Tests
The women in the screening group were posted venipuncture equipment and asked to visit their general practitioners for venipuncture. Blood samples were sent to the Ovarian Cancer Screening Unit by first class mail, centrifuged at 4,000 rpm for 10 minutes and the serum separated, aliquoted, and stored at –20°C. All blood samples received more than 56 hours after venipuncture were discarded and repeat samples requested. Serum CA125 levels were determined by commercial radioimmunoassay (CA 125II kit, Centocor, Malvern, PA) within 2 weeks of receipt of the blood sample. The CA125 values were entered into the trial database, which calculated the ROC (eg, a risk of 2% implied a risk of one in 50 of ovarian cancer), and then generated the appropriate letter.
TVS was performed either at the Ovarian Cancer Screening Unit or at a collaborating center by a consultant radiologist or experienced ultrasonographer. Where TVS was not possible or acceptable, a transabdominal scan was performed. Ovarian morphology and dimensions were assessed and volume determined using the formula for an ovoid (d1 x d2 x d3 x 0.532). Ovarian morphology was classified as normal if the ovary was of uniform hypoechogenicity and smooth outline, simple cyst if a single, thin walled, anechoic cyst with no septa or papillary projections was detected, and complex if there was nonuniform echogenicity because of cystic and solid areas in the ovary, multiple cysts, or ascites. The larger of the two ovaries with the most abnormal morphology was used to classify scans as shown in Table 1.
Screening Strategy
Level I screen (CA125). Women in the screen arm underwent an annual serum CA125 assay interpreted using the ROC calculation. On the basis of the risk value, patients were allocated to one of three groups and managed as detailed below:
Normal risk (< one in 2,000). Patients were informed that their results were normal.
Intermediate risk (one in 2,000 to one in 500). Patients were recalled for a repeat venipuncture. The interval of recall varied between 6 weeks and 6 months and was inversely related to the risk estimate. Management following repeat testing depended on the recalculated risk value, which incorporated the latest CA125 result.
Elevated risk (> one in 500). Patients were recalled for a level II screen.
Level II screen (TVS and CA125). Women underwent a scan of their ovaries and serum CA125 assay. On the basis of the results of these tests, they were managed as presented in Table 2.
Normal scan and risk less than one in 25. Patients were informed that their results were normal.
Normal scan and risk one in 25 to one in five. Serum CA125 was repeated and the risk reassessed. Subsequent management was determined by the same risk criteria as described above for level 1 screens.
Equivocal or unsatisfactory scans irrespective of the risk value or normal scan with risk more than one in five. TVS and CA125 were repeated after ruling out other conditions associated with a CA125 elevation. The patient was referred for surgery if the scan findings were persistently equivocal or became abnormal.
Abnormal scans irrespective of the risk value. The patient was referred to a gynecologic oncologist for assessment and possible surgical investigation.
Confirmation of Diagnosis
Operative notes and histology reports of those who underwent surgery were reviewed to confirm diagnosis and the stage, grade, and histology of any cancer diagnosed. Further information regarding the women diagnosed with ovarian cancers was obtained from the surgeon, medical oncologist, and general practitioner.
Follow-Up
All volunteers who did not undergo surgery or had not withdrawn from the study were contacted a year later for further screening and were sent a postal follow-up questionnaire 3 years after recruitment.
Analysis
The end point of the prevalence screen was defined as surgery, following referral to a specialist or return to annual screening. The prevalence screen, therefore, consisted of a single or series of serum CA125/scans culminating in surgery or return to annual screening. Index cancers were defined as primary invasive epithelial carcinomas of the ovary and fallopian tube. For the purpose of this analysis, all patients have been censored 1 year after the date of the final blood or scan test related to their prevalence screen. Their status, with regard to a diagnosis of an index cancer, was determined on the date of censorship.
Role of the Funding Agencies
The charities that funded the study had no role in study design, collection, analysis, interpretation of data, writing of the report, or decision to publish.
RESULTS
A total of 13,582 women were recruited in the United Kingdom between 1995 and 2000. Fifty women provided incomplete data. Sixty six women were ineligible, the majority (51 women) because they were premenopausal. Of the remaining, 6,682 women were randomly assigned to the screen arm. Table 3 summarizes their baseline characteristics. Figure 1 summarizes the overall course of the study.
Level I Screen
After being randomly assigned to the screen group, 150 patients did not provide a blood sample because they had both ovaries removed (seven women), developed a new cancer or other illness (11 women), withdrew (46 women), or did not respond (86 women). Of the 6,532 women who had a blood test, the risk calculation was normal in 5,213 women, intermediate in 1,228 women, and elevated in 91 women. The 1,228 women with intermediate risk values (ie, one in 2,000 to one in 500) had repeat CA125 testing. Of these women, 1,131 women were returned to annual screening, 44 women withdrew, and 53 women were found to have elevated risk (Fig 1). All 144 women (91 + 53) with an elevated risk (ie, risk > one in 500) were offered a level II screen. If a fixed cutoff of CA125 more than 30 U/mL had been used, 413 women would have required a level II screen.
Level II Screen and Further Management
Six of the 144 women in the elevated risk group were found to have nonovarian malignancies—myeloma, hepatocellular carcinoma, colon, breast, endometrial cancer, and adenocarcinoma of unknown primary (ACUP). In the latter case the woman had a risk estimate of more than one in five with normal ovaries on scan. On computed tomography, she had ascites with a large splenic mass, extensive upper abdominal lymphadenopathy, a 2 cm right external iliac mass, and masses in the upper body of stomach and lower lobe of right lung but no pelvic abnormality. Biopsy of a lymph node confirmed metastatic adenocarcinoma and the diagnosis on multidisciplinary review was of abdominal carcinomatosis of unknown primary origin.
The results of the level II screen in the remaining 138 women were classified as shown in Table 4 with the overall outcome summarized in Figure 2.
Normal scan and risk less than one in 25. All 39 patients in this group returned to annual screening.
Normal scan and risk one in 25 to one in five. All 28 women had a further CA125 and were then returned to annual screening.
Equivocal or unsatisfactory scans irrespective of the risk value or normal scan with risk more than one in five. All 58 women had a repeat CA125 and TVS. Six women underwent surgery as the repeat TVS was abnormal (two women) or persistently equivocal (four women). Fifty-two women returned to annual screening.
Abnormal scans irrespective of the risk value. Thirteen women were assessed clinically and 10 women underwent surgery. Three women (risk values < 1%) were managed conservatively; one had a diagnosis of uterine fibroids, a second had a biloculated small ovarian cyst with no sinister features, and the third had multicystic ovaries with the diameter of the largest cyst less than 2 cm. The latter was followed-up with repeat scans and bloods for 6 months during which time the ovarian size decreased.
Surgery in Screen Positives
Overall, 16 women underwent surgery as a result of prevalence screening. Eleven women had surgery at Bart's Hospital, London and five women had surgery in other hospitals after referral by their general practitioner. Seven women had a laparotomy, four women a diagnostic laparoscopy, and five women a laparoscopic salpingo-oophorectomy. Two of the latter women had completion laparotomy as they were found to have ovarian cancer. Postoperatively, one woman (without ovarian cancer) developed bowel obstruction and had to undergo a further laparotomy, with drainage of a pelvic abscess and resection of small bowel, caecum, and ascending colon.
Final Diagnosis in Screen Positives
Ovarian malignancy was detected in five women. One had ovarian recurrence of a breast cancer 16 years after initial diagnosis, one had a borderline papillary serous ovarian carcinoma, and three had primary invasive epithelial ovarian cancers (EOCs) (Table 5). The details of the benign findings in the remaining 11 women are summarized in Table 6.
Follow-Up
Follow-up data at 1 year after the final test related to the prevalence screen was available in 97.5% of the screened women (6,369 women). No additional cases of ovarian cancer were documented during follow-up.
Screening Performance
The screening strategy achieved a PPV of 19% (Table 7) for detection of primary invasive EOC. Specificity was 99.8%, if all participants are included. If only those individuals who were followed for at least 1 year are used to calculate specificity, then it was 99.7%.
DISCUSSION
This report describes the first prospective use of the ROC algorithm in screening for ovarian cancer. The results demonstrate that OCS, using the algorithm is feasible and can achieve high specificity (99.8%; 95% CI, 99.7% to 99.9%) and PPV (19%; 95% CI, 4.1% to 45.6%). The initial algorithm was developed using stored serum from apparently healthy women in the Stockholm OCS study, six of whom were found to have ovarian cancer.8 Retrospective analysis estimated it achieved a specificity of 99.7% and PPV of 16% for detection of ovarian cancers diagnosed within a year of the assay.5 The algorithm was then applied retrospectively to 33,621 CA125 results from 9,233 women from the Bart's OCS trials. When compared with a fixed cutoff for CA125, the area under the curve significantly improved from 84% to 93% (P = .01).4 The results reported here demonstrate for the first time in a prospective study that this novel screening strategy algorithm can achieve the high specificity and PPV required for OCS. These are essential characteristics as positive screens result in surgical investigation, with its attendant morbidity.
The trial was undertaken to inform the design of a definitive OCS trial with mortality as a primary end point and to assess the feasibility and logistic implications of this approach. On the basis of the initial blood test, 18.8% of women had an intermediate risk value. On repeat testing, 92% with an intermediate value were recalculated to be at normal risk and returned to annual screening. The initial protocol did not specify the number of repeat tests, and 252 women were referred for four or more tests (Fig 1). Of the women recalled for four or more tests, 6% (15 out of 252) withdrew from the study compared with 1.5% (18 out of 1,228) who withdrew after their first intermediate risk result (Fig 1). It was noted that the 15 women who withdrew after multiple tests gave the disruption and stress resulting from repeat testing as their primary reason for withdrawing. For the definitive UKCTOCS trial it was, therefore, decided that a maximum of three CA125 tests would be requested, and that women who remained in the intermediate group would then be offered an ultrasound scan. Assessment of anxiety and worry resulting from repeat testing is one of the core objectives of the psychosocial study that accompanies UKCTOCS.
The overall rate of referral for ultrasound assessment of the ovaries was 2.2% (144 of 6,514) using the ROC algorithm, compared with a rate of 6.3% (413 of 6,514), had a fixed cutoff for CA125 (> 30 U/mL) been used. Ultrasound is an expensive and labor intensive procedure, and the higher rate would have a significant impact on cost and logistics of a definitive trial and, if successful, a national screening program.
In the course of the study, six nonovarian cancers were diagnosed as a result of elevated CA125 levels. These included a case of ACUP. The findings on imaging in this particular individual made it unlikely that the primary site was ovarian. However, the classification of ACUP raises a difficult issue in OCS trials. Ovarian cancer, if diagnosed at an advanced stage, may wrongly be classified as ACUP, resulting in lower numbers of ovarian cancer cases (true-positives). This can cause under reporting of ovarian cancer cases in the control arm of a screening trial. It is important that as much information as possible be gathered about such cases and that the number of ACUP cases in the screen and controlled arms are monitored; the numbers should not be significantly different.
Sixteen operations were undertaken following the level II screen and clinical assessment. Eleven patients did not have ovarian malignancy. The major cause for false-positive surgery was benign ovarian cystadenomas and the presence of pelvic adhesions. In two cases this was related to diverticulitis, which is a known cause for raised CA125 levels. Initial laparoscopic assessment made it possible to avoid oophorectomy in women with normal ovaries but dense pelvic adhesions, where the risk of surgical complications may be increased.
Of the five women found to have ovarian malignancy, one had a stage IA borderline serous micropapillary ovarian cancer. There is controversy among trialists about inclusion of borderline tumors (tumors of low malignant potential) as true-positives in OCS trials. Recently, it has been proposed that the serous borderline tumors should be classified as micropapillary serous carcinoma or atypical proliferative serous tumor. The former have a higher incidence of bilateral ovarian involvement, recurrence with shorter progression-free interval, and invasive implants. However, the overall survival of patients in both groups appear to be the same,9 and both these tumors have an overall good prognosis. We have, therefore, continued our practice of not including borderline ovarian carcinomas as true-positives irrespective of their histology as the aim is to develop a strategy that impacts on ovarian cancer mortality.
Ovarian recurrence of a primary breast cancer was detected in one patient. Women who had previously been diagnosed with a malignancy and were in remission were allowed to join the study. A total of 5.12% (689 women) in the screened arm had a personal history of cancer, mainly breast and bowel. As CA125 can occasionally be elevated in recurrences of such malignancies,10 the detection of occasional asymptomatic recurrences is inherent in the study design. The benefits of such detection are uncertain and women with a previous history of cancer should be counseled about this at recruitment.
The prevalence screen reported here involved 6,532 women. Three women were diagnosed with primary invasive EOC, two women stage I, and one woman stage II. No other primary ovarian cancers were reported in these women during the year following their screen. The numbers are too small to comment on sensitivity or stage shift but are consistent with the data from retrospective evaluation of the ROC algorithm. There is preliminary evidence from our previous RCT that early detection may increase long-term survival.3 Follow-up data from prospective single-arm OCS trials also provide evidence of a possible survival benefit.11,12 It is, however, important to note that the ovarian cancers detected by screening in the latter trials included granulosa cell and borderline ovarian carcinomas that have a good overall prognosis.
Large RCTs are required to assess the impact of screening on ovarian cancer mortality in the general population. In the United States the Prostate, Lung, Colorectal, and Ovarian Cancer study involving 74,000 women includes an OCS arm.13-19 The trial reported here has informed the design of the UKCTOCS, which aims to randomly assign 200,000 postmenopausal women, aged 50 to 74 years, in a 1:1:2 ratio to annual ultrasound screening, CA125 screening using the ROC algorithm, and a control group (www.ukctocs.org.uk). UKCTOCS includes a comprehensive evaluation of cost, morbidity, compliance, and acceptability of screening, in addition to the primary end point of ovarian cancer mortality. More than 95,000 women have now been randomly assigned into this study, which will be completed in 2011 and is designed to provide definitive data about the impact of population screening for ovarian cancer.
The algorithm is also being used for screening high-risk women in the US Cancer Genetics Network trial, and in the screening arm of the Gynecology Oncology Group study of women at high risk in the United States. In the UK Familial Ovarian Cancer Screening Study, efforts are underway to revise the current screening strategy to incorporate the ROC algorithm. If screening were shown to be effective in these various trials, it is highly likely that the ROC algorithm that will be part of future OCS programs across the world.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank all the women who participated in the trial. Also our thanks to all the general practitioners whose collaboration was essential to the study.
NOTES
Supported by grants from the UK charities Research into Ovarian Cancer and the Gynecology Cancer Research Fund.
Presented in part at the Annual Meeting of the European Society of Gynecological Oncology, Brussels, Belgium, May 2003.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Jacobs I, Davies AP, Bridges J, et al: Prevalence screening for ovarian cancer in postmenopausal women by CA 125 measurement and ultrasonography. BMJ 306:1030-1034, 1993
Jacobs I, Stabile I, Bridges J, et al: Multimodal approach to screening for ovarian cancer. Lancet 1:268-271, 1988
Jacobs IJ, Skates SJ, MacDonald N, et al: Screening for ovarian cancer: A pilot randomized controlled trial. Lancet 353:1207-1210, 1999
Skates SJ, Menon U, MacDonald N, et al: Calculation of the risk of ovarian cancer from serial CA-125 values for preclinical detection in postmenopausal women. J Clin Oncol 21:206-210, 2003
Skates SJ, Pauler DK, Jacobs IJ: Screening based on the risk of cancer calculation from Bayesian hierarchical change point and mixture models of longitudinal markers. J Am Stat Assoc 96:429-435, 2001
Menon U, Talaat A, Jeyarajah AR, et al: Ultrasound assessment of ovarian cancer risk in postmenopausal women with CA125 elevation. Br J Cancer 80:1644-1647, 1999
Menon U, Talaat A, Rosenthal AN, et al: Performance of ultrasound as a second line test to serum CA125 in ovarian cancer screening. BJOG 107:165-169, 2000
Einhorn N, Sjovall K, Knapp RC, et al: Prospective evaluation of serum CA 125 levels for early detection of ovarian cancer. Obstet Gynecol 80:14-18, 1992
Deavers MT, Gershenson DM, Tortolero-Luna G, et al: Micropapillary and cribriform patterns in ovarian serous tumors of low malignant potential: A study of 99 advanced stage cases. Am J Surg Pathol 26:1129-1141, 2002
Sjovall K, Nilsson B, Einhorn N: The significance of serum CA 125 elevation in malignant and nonmalignant diseases. Gynecol Oncol 85:175-178, 2002
van Nagell JR Jr, DePriest PD, Reedy MB, et al: The efficacy of transvaginal sonographic screening in asymptomatic women at risk for ovarian cancer. Gynecol Oncol 77:350-356, 2000
Sato S, Yokoyama Y, Sakamoto T, et al: Usefulness of mass screening for ovarian carcinoma using transvaginal ultrasonography. Cancer 89:582-588, 2000
Gohagan JK, Prorok PC, Hayes RB, et al: The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: History, organization, and status. Control Clin Trials 21:251S-272S, 2000
Hartge P, Hayes R, Sherman M, et al: Complex ovarian cysts in postmenopausal women are not associated with ovarian cancer risk factors. Preliminary data from the PLCO cancer screening trial. Ann Epidemiol 10:465, 2000
Hasson MA, Fagerstrom RM, Kahane DC, et al: Design and evolution of the data management systems in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:329S-348S, 2000
Hayes RB, Reding D, Kopp W, et al: Etiologic and early marker studies in the prostate, lung, colorectal and ovarian (PLCO) cancer screening trial. Control Clin Trials 21:349S-355S, 2000
Miller AB, Yurgalevitch S, Weissfeld JL: Death review process in the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:400S-406S, 2000
O'Brien B, Nichaman L, Browne JE, et al: Coordination and management of a large multicenter screening trial: The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:310S-328S, 2000
Prorok PC, Andriole GL, Bresalier RS, et al: Design of the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial. Control Clin Trials 21:273S-309S, 2000(Usha Menon, Steven J. Ska)