Sage Journals: Discover world-class research

Abstract

Ovarian cancer can be treated with a very good prognosis if detected in the early stages, but not after it has advanced. Transvaginal ultrasound is capable of identifying changes in ovarian size and structure, and thereby detects early ovarian malignancies. This view has generated four major trials on transvaginal ultrasound detection: the Kentucky, PLCO, UKCTOCS, and SCSOCS trials. Each is sufficiently different to warrant examination. The Kentucky, UKCTOCS and SCSOCS trials report a shift to early stage detection. The Kentucky trial reports a survival benefit, while follow-up survival analysis is pending in the UKCTOCS and SCSOCS trials. Details of these trials are presented including definitions, inclusions/exclusions, analytic structure (intention-to-treat vs per protocol), performance (sensitivity, specificity, positive predictive value and negative predictive value), extent of screening-related treatment, time from screening to treatment, length of follow-up and survival versus mortality analysis. Questions are answered here about effectiveness, application, prevalence, cost and the potential for harm.

Keywords

morphology indexing ovary screening serial transvaginal ultrasound

Medscape: Continuing Medical Education Online

This activity has been planned and implemented in accordance with the Essential Areas and policies of the Accreditation Council for Continuing Medical Education through the joint sponsorship of Medscape, LLC and Future Medicine Ltd. Medscape, LLC is accredited by the ACCME to provide continuing medical education for physicians.

Medscape, LLC designates this Journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit(s)™. Physicians should claim only the credit commensurate with the extent of their participation in the activity.

All other clinicians completing this activity will be issued a certificate of participation. To participate in this journal CME activity: (1) review the learning objectives and author disclosures; (2) study the education content; (3) take the post-test with a 70% minimum passing score and complete the evaluation at www.medscape.org/journal/wh; (4) view/print certificate

Release date: 14th December 2012; Expiration date: 14th December 2013

Learning objectives

Upon completion of this activity, participants should be able to:

Assess the epidemiology and mortality risk of ovarian cancer

Distinguish the number of successful trials of transvaginal ultrasound to screen for ovarian cancer

Analyze the accuracy of transvaginal ultrasound to detect ovarian tumors and cancer

Evaluate the methods and results of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial

Financial & competing interests disclosure

CME Author

Charles P Vega, Health Sciences Clinical Professor; Residency Director, Department of Family Medicine, University of California, Irvine, CA, USA

Disclosure: Charles P Vega has disclosed no relevant financial relationships.

Financial & competing interests disclosure (cont.)

Edward J Pavlik, Division of Gynecologic Oncology, Department of Obstetrics & Gynecology, University of Kentucky Chandler Medical Center, 800 Rose Street, Lexington, KY 40536-0293, USA

Disclosure: This work was supported by grants from the Telford Foundation, and the Department of Health and Human Services, Commonwealth of Kentucky (KY, USA).

John R van Nagell Jr, Division of Gynecologic Oncology, Department of Obstetrics & Gynecology, University of Kentucky Chandler Medical Center, 800 Rose Street, Lexington, KY 40536-0293, USA

Disclosure: This work was supported by grants from the Telford Foundation, and the Department of Health and Human Services, Commonwealth of Kentucky (KY, USA).

Editor

Elisa Manzotti, Publisher, Future Science Group

Disclosure: Elisa Manzotti has disclosed no relevant financial relationships.

Screening involves the examination of healthy individuals before any signs of disease become evident. By contrast, a diagnostic workup is appropriate whenever signs of disease are present in order to determine cause. Early stage ovarian cancers occur predominately without specific symptoms [1,2] and if discovered are highly curable with surgery and conventional chemotherapy, in contrast to late stage disease, which carries a poor prognosis despite radical tumor debulking and multiple courses of combination chemotherapy. Ovarian cancer is considered a ‘silent killer’ because there are often few signs of it until it reaches an advanced stage. Because ovarian malignancies claim more lives than all other gynecologic malignancies combined in the USA, it is imperative to try to detect this disease in an early stage of development through screening. The scope of this paper is focused on the detection of early stage ovarian tumors using transvaginal ultrasonography.

Screening

There are important tenets for screening [3,4]. Screening is appropriate for disease that is grave and threatens well-being; and curable if found early. The screening tool must be effective in finding disease that can be treated successfully and in not reporting that disease is present when it is not. It should be easy to administer and well-tolerated.

In addition, there are several considerations that are often incorporated into discussions about screening, but differ by disease and are considered here in the context of ovarian cancer.

The disease significantly impacts public health

Key to this consideration is the interpretation of ‘significantly impacts’. Diseases associated with mortality would appear to satisfy this criterion, but interpretations based on age can weight the significance in favor of earlier age risk. When prevalence is considered, it is often found to increase with age, as would disease-specific mortality. This leads to several dilemmas:

What weight should be given to significance when increased risk occurs at 50 years of age versus older ages of 60, 70 or 80 years when life expectancy is declining?

Should risk be based on the entire age population or gated for age group risk?

Is the probability of disease-specific death (as a function of all deaths or grouped deaths like cancer-specific deaths) more related to significance than prevalence?

Should the quality and cost of end-of-life experiences be included in considerations of significance in terms of avoiding extended palliative care as well as suffering and death? [5]

Should achievable cost savings be considered as part of significant impact? Significant impact: ovarian cancer kills more women than all other gynecologic malignancies combined and is associated with vastly lower costs of care in early stages than advanced stages [6] where outcomes are poor and suffering prolonged.

The disease has a high-to-intermediate probability of occurrence in the population

High and intermediate probabilities are very subjective and are often used in the absence of evidence-based findings. Ovarian cancer is one of the top five causes of cancer death for US women [7]. Over 15,500 US women will lose their lives to the disease this year and 22,280 will learn that they have ovarian cancer [7]. By way of comparison, 229,060 women will be diagnosed with breast cancer and 39,510 will die from it, showing that ovarian cancer is ten-times less prevalent, but it will kill as many as 40% of the number dying of breast cancer [7]. Thus, the expected mortality ratio for ovarian cancer (15,500:22,280 = 69.6%) is fourfold greater than the expected mortality ratio for breast cancer (39,510:229,060 = 17.2%).

The probability of occurrence and prevalence is dwarfed by the high mortality rate of ovarian cancer.

A critical point for disease detection must exist

Postulate that the screened patient is asymptomatic before a critical point, that the critical point occurs before clinical diagnosis and that diagnosis would not otherwise occur early before the critical point. Importantly, the critical point must occur early enough to achieve a favorable outcome or cure. Because treatment of lung cancer cannot be started early enough to achieve a cure, the critical point occurs too early, while it occurs too late for endometrial cancer, since abnormal uterine bleeding provides a natural early warning system.

Ovarian cancer meets all of the critical point criteria.

Medical care is available if screening is positive

In a global sense, surgery is widely available, as is chemotherapy. On an individual basis, surgery for gynecologic cancer is elective and each case takes its place in the operating room queue. We stick to the position that delays should be kept to a minimum in order to keep ovarian disease from progressing to a stage that is less treatable [8]. Thus, it is imperative that screening sites have active roles in facilitating the scheduling of surgery. Availability of surgery and chemotherapy (surgery alone for low grade stage IA malignancy or surgery and chemotherapy for all else) for women in rural areas presents a challenge. Having surgery far from home and returning for chemotherapy is difficult.

Effective treatment and medical care exists for women with early stage ovarian cancer detected by screening with women in rural areas facing a burden of traveling to access medical care.

Patients are willing & able to undergo further evaluation & treatment

Since screening has no value if there is not subsequent diagnosis followed by treatment, it is important that encouraging support facilitates moving forward. For the woman without medical insurance this facilitation includes financial counseling.

Our extensive experience with over 37,000 women who received over 200,000 screens indicates that it has been possible to treat all women that had a transvaginal ultrasound (TVS) screen indicative of malignancy, regardless of having medical insurance.

Ultrasonography

The first efforts to identify ovarian tumors using ultrasound were reported by Campbell et al. [9]. These efforts employed abdominal ultrasound, but this abdominal approach was less accurate in defining ovarian details because the distance from the probe to the ovaries is large and includes significant airspace in which soundwaves attenuate. In TVS, the probe has a closer proximity to pelvic structures since most of the pelvic anatomy of interest is within 9 cm of the vaginal fornices. TVS use originated with fertility medicine in the mid-1980s [10 –12], and was adapted to imaging ovarian pathology before the decade ended [13,14]. TVS with Doppler blood flow imaging and power Doppler is used to evaluate blood flow to ovarian tumors and to identify patterns of flow associated with ovarian neoplasia.

Factors for evaluating tumor detection performance

Factors used for evaluation are: true positives (TP; surgically confirmed malignancy after a positive screening test), true negatives (TN; no malignancy reported during the 365 days following a negative screen), false positives (FP; no malignancy at surgery after a positive screening test) and false negatives or interval cancers (FN; malignancy diagnosed within 365 days of a negative screen). Sensitivity is calculated to determine performance when disease is present (TP/[TP + FN]), while specificity is estimated to determine performance when the disease is absent (TN/[TN + FP]). The positive predictive value (PPV) is calculated to determine how many of those that test positive (TP/[TP + FP]) really have ovarian cancer, while the negative predictive value is calculated to determine how many that test negative really do not have ovarian cancer (TN/[TN + FN]). The greater these estimates are, the higher is tumor detection performance. Ovarian tumors of low malignant potential (LMP) are biologically less aggressive than invasive ovarian malignancies, but are counted as TPs because of their potential to spread beyond the ovary. LMPs rarely recur as invasive cancer, but those having micropapillary projections and/or those associated with peritoneal implants present a greater risk to women [15,16]. Tumor progression comprises a continuum that includes ovarian tumors of LMP [17] and indicates the potential involvement of LMPs in malignant progression, especially since K-RAS or B-RAF mutations of low grade serous carcinoma are also found in serous ovarian tumors of LMP [15]. Further supporting the continuum of progression is the finding that ~60% of low grade serous carcinomas contain areas of serous tumor of LMP [18]. Moreover, a serous ovarian tumor of LMP with extreme architectural complexity can be difficult to distinguish from a low-grade serous carcinoma resulting in significant interobserver variation among histopathologists in clinical classifications [15]. Large mucinous ovarian tumors of LMP may be difficult to adequately sample, so there is a risk of missing some small focus of malignancy [15]. Importantly, serous and mucinous ovarian tumors of LMP can be lethal [16,19]. Fallopian tube malignancies are considered as ovarian malignancies in three trials summarized here, while peritoneal cancers are excluded because they are not visualized by TVS. This exclusion has little impact on performance estimates because peritoneal cancer incidence is 5.7% that of ovarian cancer [20]. Thus, the approach in these trials is to include ovarian tumors that present a risk of malignancy and to exclude from performance evaluations malignancies that cannot be detected by TVS.

Evaluating the effect of ovarian tumor detection on survival

The evaluations of tumor detection performance can be made at any time point; however, the effect of this detection on disease-specific survival needs to be made only after sufficient time has passed for mortalities to have occurred. Estimates should include the survival of all ovarian malignancies in the screened group (i.e., both TPs and FNs). We have taken the position that ovarian tumors of LMP should be excluded from survival analysis so that the included cases reflect the effects on ovarian malignancies with the highest mortality risk. In the context of survival analysis, the comparison or control group warrants some consideration. There are two alternatives that must be considered. First, the survival of all screened cancers should be compared with the survival of all unscreened cancers representing the same region and time frame as those in the screened group. In the second alternative, all individuals in the screening group should be compared with all individuals in a group that hasn't been screened. Because there will be a relatively small number of cancers compared with a very large fraction of women without cancer, it will be difficult to detect significant differences between the groups, making the second alternative less favorable. It is best to use life tables from the Kaplan–Meier analyses to obtain numeric values of survival for screened subjects and unscreened members of the comparison groups, including error estimates, while overall significance for the effect of ovarian tumor detection on survival can be decided through Kaplan–Meier analysis that compares the screened and unscreened comparison group. Finally, in order to be sure that unknown ovarian cancer deaths do not exist, all participants must be checked for living versus dead status using, for example, the Social Security Death Index and then cause of death, determined through vital statistics and hospital records. Evidence of disease is the most significant single factor for assigning death due to ovarian cancer, especially since older members may succumb to other conditions and diseases. Consideration of a trial with participants randomized to intervention (screened) and control (unscreened) groups is appropriate. In the case of screening, the definition of the intervention group must extend beyond the initial randomized assignment to include both screening and, when deemed necessary by screening, expeditious diagnosis and treatment. While the unscreened definition of the control group may appear straightforward, this group has identities established by more than randomization to not receiving screening. For example, the identity of this group may not be the same as the screened group with regard to health conscious members who stay the course of the study, or members of the control group that cannot seek or receive ‘normal’ care. In studies where there is a low probability of the event occurring in each group (i.e., ovarian cancer), it may be difficult to achieve statistical significance. This difficulty can be circumvented by using a population control group where its large numbers can make significant contributions to the power of statistical comparisons. While randomized controlled trials are very powerful, there are settings that challenge them and ovarian cancer screening can be one of these settings.

Framework & findings: results from trials using ultrasound for the early detection of ovarian tumors

There are four published trials that summarize findings on the use of ultrasound for the early detection of ovarian tumors: the Kentucky trial [21], the PLCO trial [22], the UKCTOCS [23] and the SCSOCS trial [24]. It should be noted that a separate UKFOCSS for high-risk women is underway in the UK. These trials have some similarities, but also have significant differences and, accordingly, are deconstructed and summarized in Table 1. The Kentucky trial is ongoing and uses population control groups, while the others have been closed and used randomized control groups. Intention-to-treat analysis was employed by the PLCO, UKCTOCS and SCSOCS trials, which is based on the initial screening treatment intent (not on whether screening was actually administered), and uses data from all subjects randomized to the screening group, including those who did not receive screening or treatment indicated by screening. Per protocol analysis was employed in the Kentucky trial in which only subjects who complete screening and treatment (as indicated by screening protocol) are counted towards the final results. In the Kentucky trial, all subjects were screened, and all received surgery and treatment when their screening results indicated that this was appropriate relative to the trial protocol. All of the trials were gated on age or postmenopausal status; however, the Kentucky trial also accepted high-risk women 25 years or older with a family history of ovarian cancer in a primary relative. Results of only the prevalence screen (i.e., initial screen) are reported for the UKCTOCS and SCSOCS trials, while results from multiple rounds of screening are included in the Kentucky and PLCO trial reports. The Kentucky trial and one arm of the UKCTOCS trial employed screening with TVS alone, while TVS combined with CA-125 was used in the PLCO and SCSOCS trials. Some women in the Kentucky trial who had an abnormal TVS screen were subsequently tested with CA-125 (n = 1284). The last two screening rounds of the PLCO trial used CA-125 alone. The UKCTOCS trial included the most women who received TVS screens (42,416), while the Kentucky trial administered the most screens (205,190, almost ~100,000 more than the next highest [108,280 in the PLCO trial]). Because primary peritoneal cancers cannot be visualized with TVS, they were excluded from analyses in the Kentucky, UKCTOCS and SCSOCS trials, but not from the PLCO trial. We define a shift to early stage disease to be greater than the percentage that occurs in the unscreened group. The biggest shifts to the detection of early stage I and II epithelial ovarian cancer were in the Kentucky trial (70.2 stage I and II vs 27% in the unscreened comparison population control group) and in the SCSOCS (67%) and UKCTOCS (50%) trials, while the PLCO trial (27.8%) failed to detect a shift to early stage disease. A majority of aggressive type II ovarian cancers were detected in each trial, indicating that TVS screening can detect both the most pernicious (type II) ovarian cancers as well as the more indolent type I ovarian cancers. Epithelial ovarian tumors of LMP were counted as TPs in the Kentucky and UKCTOCS trials. Follow-up of TP epithelial ovarian cancer was up to 14.8 years in the Kentucky trial, 14 years in the SCSOCS trial, 5.54 years in the UKCTOCS trial and unreported in the PLCO trial. Data for TP, FP, TN and FN were present in reports from the Kentucky, UKCTOCS and SCSOCS trials so that sensitivity, specificity, PPV and negative predictive value could be determined; however, this was not possible with the PLCO trial report. The highest overall TVS sensitivity was reported in the UKCTOCS trial (63/[63 + 8] = 88.7%) and remained high when limited to 45 primary ovarian and fallopian tube TPs (84.9%). Sensitivities were almost as high in both the Kentucky (86.4%) and SCSOCS (79.5%) trials, indicating the TVS screening is effective in finding malignancies. Specificities were very high, ranging from 98.2–99.9% for the Kentucky, UKCTOCS & SCSOCS trials, indicating TVS screening is extremely effective at showing the absence of disease. Similarly, the negative predictive values (NPV) were also very high (>99.9%) indicating that in all three trials almost all cases called negative by TVS screening were negative. The PPV indicates the nonmalignant ovarian abnormalities that TVS screening also discovers and ranged from 5.4–47.3% in the different trials. Thus, four women that didn't have a malignancy went to surgery in the Kentucky trial for each malignancy that TVS screening identified (PPV = 20.2%). The number of women was lower in the SCSOCS trial (~1 nonmalignancy to surgery for each malignancy [PPV = 47.3%]) and much higher in the UKCTOCS trial (~12–19 nonmalignancies to surgery for each malignancy [PPV = 5.4–7.5%]). An earlier report on the first four screening rounds of the PLCO trial has indicated that 19.5 nonmalignancies went to surgery for each malignancy [25]. Lastly, the Kentucky trial found a significant survival benefit to TVS screening, while the PLCO trial reported similar mortalities in the screened and control group. Postscreening follow-up is still underway in the UKCTOCS and SCSOCS trials and survival analysis is pending.

Table 1.

Summary deconstruction of trials using ultrasound for the early detection of ovarian tumors.

	Kentucky trial [21]	PLCO trial [22]	UKCTOCS trial [23]	SCSOCS trial [24]
Country	USA	USA	UK	Japan
Design	Population control	Randomized control	Randomized control	Randomized control
Status	Ongoing	Closed	Closed	Closed
Accrual	Ongoing	November 1993–July 2001	April 2001–October 2005	September 1985–1999
Follow-up	Ongoing	Up to 13 years from enrollment	Up to 13.8 years from enrollment	Up to 17.3 years from enrollment
Duration of accrual	14 December 1987–15 July 2011 (23.5 years)	1993–2001 (7.5 years)	2001–2005 (4.6 years)	1985–1999 (14.3 years)
Follow-up	Up to 23 years	Up to 13 years	Up to 13.8 years	Up to 17 years
Eligibility	≥50 years or ≥25 with family history	55–74 years	50–74 years and postmenopausal status	Asymptomatic postmenopausal women
Group screened (n)	37, 293	34,253	48,230	41,688
Total screens	205,190	150,598	48,230	41,688
TVS screens	205,190	0	42,416	0
TVS and CA-125 screens	1284 women & 5433 screens†	108,290	0	41,688
CA-125 screens alone	0	42,308	0	0
TP	76	212 (includes 29 primary peritoneal)	63 (45)	35
EOC	47	169	24	27
Stage I (EOC)	47% (22/47)	19% (32/169)	42% (10/24)	63% (17/27)
Stage II (EOC)	23% (11/47)	9% (15/169)	8.3% (2/24)	4% (1/27)
Stage I and II (EOC)	70.2% (33/47) vs 27% comparison group	27.8% (47/169)	50% (12/24)	67% (18/27)
Stage III (EOC)	30% (14/47) vs 46% comparison group	71% (120/169)	42% (10/24)	26% (7/27)
Stage IV (EOC)	0	25% (43/169)	8.3% (2/24)	7% (2/27)
Shift to early stages detected	Yes	No	Yes	Yes
Type I – indolent (grade 1)	21% (10/47)	7% (12/173 assessed)	11% (2/18 accessible)	42% (11/26 accessible)
Type II – aggressive (grade 2 and 3)	79% (37/47)	93% (161/173 assessable)	89% (14/18 assessable)	62% (15/26)
Nonepithelial ovarian cancer	5	0 or unreported	1	10
Epithelial ovarian tumor of LMP	15	21 (counted as FP)	20	Not reported
Metastatic cancer to ovary	9	0 or unreported	5	Not reported
Follow-up of EOC TP	Mean = 6.2 years, Median = 6 years (range: 0.2–14.8)	Not reported	Median = 4.57 years (range: 3.68–5.54)	Mean = 9.2 years (range: 3–14)
Surgical FP	447 (6.9% of women with TVS abnormal)	1080 (33% of 3285 women)	781	39
TN	36758	Not reported	41,564	41,613
FN	12	Not reported	8	9
Sensitivity	86.4%	Not reported	88.7%‡ (84.9%)§	79.5%
Specificity	98.8%	Not reported	98.2%‡	99.9%
PPV	14.5–20.2%	Not reported	7.5%‡ (5.4%)§	47.3%
NPV	99.97%	Not reported	99.98%	99.97%
5-year survival TP (with FNs)	84.6 ± 5.6% (74.8 ± 6.6%)	Not reported	Prevalence screen	Not reported
5-year survival comparison group	53.7 ± 2.3% at University of Kentucky Hospital (KY, USA); 48.4 ± 1.2% entire state of Kentucky	Not reported	Prevalence screen	Not reported
Stage I survival – 5 year	95 ± 4.8%	Not reported	Prevalence screen	Not reported
Stage II survival – 5 year	77.1 ± 14.5%	Not reported	Prevalence screen	Not reported
Stage III survival – 5 year	76.2 ± 12.1%	Not reported	Prevalence screen	Not reported
Survival benefit	Yes	No	Analysis pending	Analysis pending
Control group	Yes	Yes	Yes	Yes

†

Subset group that received both TVS and CA-125.

‡

Based on TP of 63 for all detected malignancies.

Based on TP of 45 for primary ovarian and fallopian tube malignancies.

CA-125: Cancer antigen 125; EOC: Epithelial ovarian cancer; FN: False negative; FP: False positive; LMP: Low malignant potential; NPV: Negative predictive value

PPV: Positive predictive value; TN: True negative; TP: True positive; TVS: Transvaginal ultrasound.

Together, the Kentucky, UKCTOCS and SCSOCS trials seemed to be more similar in their findings while the PLCO seemed more discordant. Because differences are reported in the PLCO trial with respect to the detection of early stage disease and to extended/improved survival, discussion here is focused at understanding the PLCO trial.

The PLCO trial

In the screening arm of the PLCO Trial, 39,105 women were enrolled with recruiting at different times over an eight year period ( Figure 1 ; light and medium shading) to receive four annual screens using a TVS/CA-125 combination and two annual screens using CA-125 alone so that 34,253 were available for analysis after excluding 4852 women that had received a bilateral oophorectomy. Follow-up was for a maximum of 13 years from the time of initial recruitment ( Figure 1 ; bold outline and dark shaded). Women in six start years (1993–1998) could have received 7 years of postscreening follow-up, while those in three start years (1999, 2000 and 2001) were followed for less time (6, 5 and 4 years, respectively). The PLCO report notes that women received four, five or six screening rounds depending on when they were entered into the trial. The number of women undergoing screening in each of the five rounds after the initial screening round (28,745 [T0]) ranged from 20,115 (T4) to 27,540 (T1); however, 34,253 women were included in the reported analysis indicating that between 5,508 and 14,138 women may not have had all six rounds of screening (i.e., 34,253 − 28,745 = 5508 and 34,253 − 20,115 = 14,138). The PLCO report notes that not all screen-detected ovarian cancers were treated and that for the usual care control group, a ‘high proportion’ of women underwent bimanual physical examination, although this was stopped in 1998 in project year 6, out of the 14 years that physical examinations were possible.

Figure 1.

Schematic of the PLCO Trial. Women had four, five or six screening rounds, depending on when they were randomized. CA-125: Cancer antigen 125; TVS: Transvaginal ultrasound.

PLCO survival analysis

The PLCO Trial reported 212 ovarian cancer cases observed during the screening portion and during the postscreening follow-up portion ( Figure 1 ; shown in dark shading), which for six of the nine entry years lasted 7 years after the last intervention, and for the remaining three entry years was 6, 5 or 4 years long. Reading off the cumulative incidence in Figure 2 of the PLCO report [22], it seems that approximately half of the 212 cases used in the survival analyses occurred during the 7 years of follow-up (or during the follow-up that was 4–6 years in length). Using Table 3 of the PLCO report [22], the screen-detected cases that occurred during postscreening follow-up calculates at 41% (86 out of 212). The problem with this analysis is that it presumes that screening confers an extended multiyear protection against malignancy. In the extreme case of someone who only had the initial screen, this presumed protection would be for 13 years and for someone who had all six screens the protection would be presumed to cover 7 years (or 6, 5 and 4 years for the women enrolled in project years 6–8; see Figure 1 ). In the case of rapidly proliferating ovarian malignancies, it is quite possible that the first events of ovarian malignancy started years after screening, in the 5th, 6th or 7th year of PLCO follow-up. In this context, where follow-up is considerably separated in time from screening, it is inappropriate to include approximately half of the incident cases in survival analyses because their existence during screening cannot be presumed. This way of following incidence seriously affects mortality estimation because it may include deaths during the follow-up period that were the result of ovarian malignancies that had their beginnings long after screening.

Table 2.

Mortality in the group screened in the Kentucky Trial.

Cause of death	n (%)
Accident	55 (2.8)
Heart disease/stroke	544 (27.7)
Respiratory insufficiency	250 (12.7)
Brain cancer	27 (1.4)
Breast cancer	46 (2.3)
Colorectal cancer	59 (3.0)
Endometrial cancer	9 (0.5)
Leukemia	28 (1.4)
Lung cancer	159 (8.1)
Lymphoma	42 (2.1)
Pancreatic cancer	44 (2.2)
Primary peritoneal cancer	18 (0.9)
Other cancers	131 (6.7)
Ovarian cancer	39 (2.0)
Interval ovarian cancers	12 (0.6)
Out of interval compliance	17 (0.9)
Detected ovarian cancers	10 (0.5)
Other causes	240 (12.2)
Dementia/neurologic disease	179 (9.1)
Suicide	12 (0.6)
Homocide	2 (0.1)
Unknown causes/no death certificate	80 (4.1)
Total deaths in screening program	1964 (100)

Table 3.

Recovered dollars and equivalent screens by preventing progression to stage III ovarian cancer.

	Collections/case	IIIC Expense [B] = 36 × [A] in 2006 dollars	IIIC Expense [C] = [B]/[c][f] in 2011 dollars	Screen equivalents [D] = [C]/$40	Actual screen equivalents adjusted for FP = 447 [E] = [D] − 78,225 screens	Target screen equivalents adjusted for PPV = 20.2% FP = 300; [F] = [D] − 52,542 screens
Stage IIIC	[A]	[B]	[C]	[D]	[E]	[F]
n	25	28	28	US$40/screen	78,225 screen equivalents	52,542 screen equivalents
[a]: mean	US$92,100	US$2,578,800	US$2,874,916	71,873	–	19,331
[b]: SEM	US$10,280	US$287,840	US$320,892	8022	–	NA
[c]: 95% level: [a] + 1.96 × [b]	US$112,249	US$3,142,966	US$3,503,864	87,597	9372	35,055
[d]: highest	US$239,600	US$6,708,800	US$7,479,153	186,979	108,754	134,437
[e]: Avastin®: Cohn et al. (2011) [41]	US$228,240	NA	US$6,595,170	164,879	86,654	112,338
[f]: total estimate ([c] + [e])	US$340,489	NA	US$10,099,035	252,476	174,251	199,934

[A]: Unit case cost based on collections as reported ([a] and [b]) for Stage IIIC cases [6]. Avastin expense alone is based on the literature [e] [41]. The total estimate for care includes Avastin ([f] = [c] + [e]). A total of 76 ovarian malignancies have been culled to the 33 early-stage epithelial ovarian malignancies detected here ([B]: 22 Stage I and 11 Stage II) and further reduced to 28, which represents eliminating five cases that would be found without screening using the percentage of Stage I and II epithelial ovarian cancer that occurs for unscreened women in the Kentucky Cancer Registry (18.8% stage I and 8.6% stage II). We take these 28 as a set representing cases that have been prevented from progressing to Stage III by screening. The cost of these cases is then calculated as if they had progressed to Stage III by multiplying unit costs in column [A] by 28. [C]: Conversion to 2011 dollars (2006 base cf = 0.897, 2010 base cf = 0.969) [101]. [D]: Screen equivalents of expenses in C based on $40/screen [39,40]. [E]: Adjustment for false positives = 447 as previously reported [21]. This adjustment is based on $7000 collections/case as 78,225 screen equivalents at $40/screen ([447 × $7000]/$40 = 78225). False positives surgical cost is estimated at a charge for BSO at ~$18800 and an average collection rate of 34%, which was rounded to $7000. [F]: Adjustment for achieving a PPV = 20.2% as reported [21] that would yield false positives = 300 at $7000/case resulting in 52,542 screen equivalents at $40/screen ([300 × $7000]/$40 = 52,542).

FP: False positive; PPV: Positive predictive value; NA: Not applicable; SEM: Standard error of the mean.

Detection reported in the PLCO trial

In Table 2 of the recent PLCO trial report [22] detection in the intervention group (i.e., the group screened) is summarized as screen-detected cases during (n = 20 + 53 = 73) and after screening (n = 78) including 37 interval cancers (i.e., false negatives) and as never-screened cases during and after screening (n = 16 + 8 = 24) for a total of 212 ovarian cancers reported. First, it is hard to reconcile including 24 never-screened cases in the results summarized for the screening group in Tables 2–4 of the PLCO report [22], as well as in survival analysis shown in PLCO Figure 2 [22]. Second, by isolating the events of the screening period from those in the follow-up period because of the follow-up cases that may have begun well after screening as described above, it is possible to estimate the sensitivity of cases identified during screening as 66.4% (TP/[TP + FN] = [20 + 53]/[20 + + 37]). This is low sensitivity and compares quite unfavorably with the other screening trials in Table 1. Of interest is the sensitivity in the first four rounds of the PLCO (TP/[TP + FN] = 60/[60 + 19] = 76%), which employed both TVS and CA-125 [25]. In the last two rounds of the PLCO trial, which used CA-125 alone [22], interval cancers increased to 9 FN/year from 4.75 FN/year in the first four rounds [25] indicating that CA-125 by itself missed more cancers. In the earlier PLCO report [25], TVS alone picked up 71.4% of stage I/II cases in contrast to 11.1% by CA-125 alone; however, the summary PLCO reports that only 22% stage I/II were detected overall [22]. A reasonable interpretation of this is that the use of a single CA-125 trigger will have much reduced ability to detect malignancies, especially early stage disease, making the inclusion of the last two screening rounds that utilized CA-125 alone very unlikely to perform well.

There are four factors that account for the failure of the PLCO trial to detect early stage ovarian cancer. First, is the use of CA-125 alone. Second, is the inclusion of never-screened participants in the intervention group. Third is the inclusion of the never-treated group of women in survival analysis of the screened group because they couldn't possibly receive a benefit of early detection that would improve their survival. Fourth, is the appearance of ovarian malignancies very long after screening in the extended follow-up period. These failures enrich the screen-detected cases with late stage disease, which will have a poorer survival. Including cases that did not receive treatment or cases that may have started in the extended postscreening follow-up period pushes the survival of the PLCO screening/intervention group lower and makes it indistinguishable from the control group.

Others too have reviewed the PLCO report and outlined the critical flaws of this screening approach. Specifically, ‘important limitations’ have been noted [26] and that the PLCO is “not the definitive answer on ovarian screening using these techniques” [27]. With regard to the mortality end point in the PLCO trial, “Basing decisions on the outcome of death ignores vital dimensions of life that are not easily qualified … There is more to life than death” [28]. Moreover, in contrast to serial TVS with morphology indexing, the single trigger approach in the PLCO trial for identifying sonographic positive screens resulted in a high rate of FPs. Although the PLCO trial constitutes an effort that failed to demonstrate benefits from the screening approach taken, we have presented evidence here from screening approaches that have been executed differently, which might benefit practitioners by showing the highly positive sensitivity and specificity results from three trials in which shifts to the detection of early stage disease occur.

Questions worthy of discussion

Is the early detection of ovarian tumors using ultrasound effective?

Improvement in survival is the best way to assess the effectiveness of the early detection of ovarian tumors using ultrasound. Three of the four trials reported a shift to the detection of early stage disease and follow-up for the effect on survival is still underway for two of these three. It is not surprising that the failure to detect early stage disease in the PLCO trial resulted in similar mortalities in the intervention and usual care groups. Importantly, analysis of survival must be realized as distinct from analysis of mortality. A disease-specific survival analysis will reveal an interval benefit of the intervention group relative to the comparison group. This interval benefit is greatest in the Kentucky trial for a 10-year interval and is followed by a constant continuing benefit in survival as shown in an updated survival analysis in Figure 2. There are two possibilities if the survival analysis in Figure 2 is extended further in time. First, the disease-specific survival benefit will persist if mortality is caused by diseases other than ovarian cancer. Second, the screened and unscreened curves will tend to converge due to women with late stage disease detected by screening when residual disease is responsible for recurrence. In addition, women with screen-detected stage IA ovarian cancer who have only one ovary removed could possibly experience an unrelated new primary ovarian cancer in the remaining ovary, which, if unnoticed, could appear as late stage disease and contribute to disease-specific mortality. In any of these cases, an interval of survival benefit should become the basis for the effectiveness of screening. This interval benefit will not be picked up in mortality analysis, especially after long-term convergence occurs.

Figure 2.

Kaplan-Meier survival of screened and unscreened women, p-values for difference between screened and unscreened groups. Survivals have been updated to August 2012.

All-cause mortality was determined and reported in the PLCO trial demonstrating 8.5% deaths in the screening group. In the Kentucky trial, 5.2% of the group screened have died from causes that are summarized in Table 2 , showing only 4.1% dying of unknown causes. Both the Kentucky trial and the PLCO trial made significant efforts to track deaths due to ovarian cancer in the groups screened so that unrecognized ovarian cancer deaths are not likely to have accounted for differences between the trials. Results from the Kentucky, UKCTOCS and SCSOCS trials indicate effectiveness in terms of a shift to early stage disease, while the improved survival in the Kentucky trial also indicates the effectiveness of early ultrasound detection of ovarian tumors. This concept of effectiveness is important in light of the evolving concept of the origin of this disease in the distal oviduct that have been coordinated with premalignant changes in gene function in the tubal mucosal [29]. Detection with TVS is effective and clearly holds benefits that can be achieved as we await the development of much anticipated detection tools that will target tiny nests of cells in the oviduct.

Are there differences in the application of an ultrasound screening algorithm that would affect screening results?

Delaying diagnosis and treatment on screen-detected cases would allow early stage disease to progress. The PLCO trial allowed up to 9 months for diagnostic evaluation [22]. In the Kentucky trial, 73% of the TPs received surgery within 30 days of their last suspicious screen (mean = 32 ± 4.5 days; median = 16 days, unreported). The number of surgical FPs was higher in the PLCO trial ( Table 1 ; n = 1080) and lower in the Kentucky trial ( Table 1 ; n = 447), probably because the PLCO trial used a single event trigger to advance a suspicious screening to surgery while the Kentucky trial utilized serial TVS screening with morphology indexing [30]. Thus, both a morphology scoring system and multiple screens provided a basis for advancing a suspicious case to surgery. Since many abnormal ovarian structures will resolve, as shown in Figure 3 , following a decreasing morphology score in serial screens provides a basis for sending fewer cases to surgery.

Figure 3.

Final view of complex structures before resolution (i.e., disappearance). (A) Unilateral abnormalities, never occurring on both sides simultaneously. (B) Unilateral abnormalities having had bilateral (synchronous) findings. (C) Bilateral findings had abnormalities simultaneously on both sides. The first number below each view is an encoded identifier, followed by screening visit number (i.e., v3) and age in parenthesis.

Is the prevalence of ovarian cancer fatal to screening?

It is often stated that the prevalence of ovarian cancer, which is approximately seven-times lower than breast cancer, presents a limitation that cannot be overcome even by very high sensitivity and specificity [31]. The Kentucky trial reports a PPV of 20.2% ( Table 1 ), which implies that four benign cases will be revealed at surgery for each malignancy. This has changed to three benign cases being revealed at surgery for each malignancy (PPV = 24.7%) in unreported Kentucky trial data for the period occurring after the data in the 2011 Kentucky report was analyzed [21]. How good is this level of performance? This performance with serial TVS and morphology indexing is as good – or better than – mammography [32], mammography assisted by sonography [33] or mammography combined with MRI [34] with regard to sensitivity, specificity, PPV [35 –37] and NPV. Moreover, a PPV of 24.7% compares favorably with PPVs for a nonscreening clinical setting (PPV: 29.5% overall, 17.9% premenopausal and 39.1% postmenopausal) in a group of 516 women that had a surgical intervention based on the presence of an ovarian tumor [38]. These comparisons argue that the detection of ovarian tumors using ultrasound performs in the region of benchmarks from clinical medicine.

It should be realized that prevalence may be used to infer that too many cancer-free cases must be screened for each malignancy so that the issue of prevalence is really a cost issue.

We have previously reported the costs of stage IIIC ovarian cancer [6]. Preventing progression is the important outcome of TVS screening, not only in terms of the survival advantage that is gained, but also in terms of cost savings. Using collections at this institution as the basis for the cost of care for stage III ovarian cancer [6] ( Table 3 column [A]), we apply this unit cost to what would have been the cost if the women with early stage ovarian cancer detected by TVS screening in the Kentucky trial had progressed to stage III ( Table 3 column [B]) and then convert this cost to 2011 dollars for 28 women, having corrected for five women (of the 33 in Table 1 women with stage I or II disease) that could be expected to have been detected without screening ( Table 3 column [C]). We have re-examined screening costs, taken the cost of a single TVS screen as reported previously [39,40], adjusted it to 2011 dollars and rounded it up to $40 per screen. A $40 per screen cost is the cost in the Kentucky trial screening center where the sonographer schedules are kept full all day. This $40 unit screen cost is then divided into the aggregate costs ( Table 3 column [C]) to yield the number of screen equivalents resulting from the cost of care for the 28 women that would have progressed to stage III without screening ( Table 3 column [D]). Next, two adjustments are made for the cost brought to screening by the surgeries on women identified as FPs. The first adjustment considers all 447 FPs reported in the Kentucky trial ( Table 3 column [E]), while the second adjustment is based on the greatest and most recent PPV reported in the Kentucky trial (20.2%; Table 3 column [F]). Since Avastin® (bevacizumab) was not yet available for 1991–2002 where our published stage IIIC survival study was focused, we have made estimates of how the addition of Avastin to stage III treatment would influence costs using a recent literature source for the cost of Avastin ( Table 3 row [e]) [41], and this is added to treatment (in Table 3 row [f]). Using estimates based on the FPs reported here (n = 447), 85% of the 205,190 screens delivered in the Kentucky trial report [21] would be offset by savings in cost of care from preventing progression to stage IIIC ( Table 3 column [E] row [f]). Furthermore, using estimates based on the improved PPV (20.2%) reported in the Kentucky trial, 97% of the 205,190 screens delivered would be offset by the cost of care due to preventing progression to stage IIIC ( Table 3 column [F] row [f]). Thus, the cost of providing screening in our data set is nearly balanced by the cost benefit gained by preventing progression. However, this balance disappears when the current Medicare reimbursement at this institution is used as the cost of screening (CPT 76856: $151.47), covering only 7% of the 205,190 screens when FP = 300. Taken together, this type of analysis indicates that strategies must be employed that keep the cost of screening under $50/screen.

Missing from the estimates presented here is the value of life lost due to cancer. The economic burden of OvCA is very high (~$28 billion/year [42]), translating to $1.8 million for each of the 15,500 US women that will succumb from OvCA in 2012 making the economic gain achieved by preventing one OvCA death enough to pay for 45,000 screens. The value of life lost is expected to increase 48.1% by 2020 [42] so that the value of lives lost continues to shift economics in favor of ovarian screening. Preventing of ovarian cancer deaths considered in terms of the value of life lost more than pays for the cost of screening, even at the Medicare reimbursement rate (CPT 76856: $151.47).

Does screening for ovarian tumors with ultrasound do more harm than good?

The title of a recent article took the position that “Screening Women for Ovarian Cancer Still Does More Harm Than Good [31],” being prompted by results in the PLCO trial report [22]. Regarding this question, neither ‘harm’ nor ‘good’ were defined or evaluated in the PLCO trial. Factors to be considered regarding this question are the ovarian screen itself, which is very well tolerated [43] and one of the things women overwhelmingly have indicated they want [44]. The risk of a suspicious TVS study leading to surgery and a FP outcome have already been discussed here, being similar to clinical benchmarks. Complications occur in surgery and can be graded with regard to their seriousness [45]. In the PLCO trial, complications were not graded and consequently the question of harm due to serious high-grade complications was not addressed. Nonphysical or psychological harm to women has been and continues to be examined with women in the Kentucky trial. When compared with an age and education matched group with no history of ovarian screening, women in the Kentucky trial had more ovarian cancer-specific distress/anxiety, less optimism and less knowledge about risk factors upon entry [46]. Thus, some distress or anxiety relative to ovarian cancer appears to play a motivating role for entering the Kentucky screening trial. As part of these efforts, the validity of self-reporting by women in the Kentucky trial was evaluated and found to be very high [47]. In a study with baseline, 2-week and 4-month measurement, recipients of a normal ovarian screening examination showed decreased ovarian cancer-related distress, increased positive effect and increased knowledge of risk factors [48], indicating that for the vast majority of women screened there are beneficial effects on ovarian cancer-specific anxiety, attitude and knowledge. Women who received an abnormal TVS screening result were found to have an elevated ovarian cancer-specific distress (but not general distress) at 2-week follow-up that returned to baseline at the 4-month follow-up [49]. Results were influenced by a monitoring coping style, low optimism and family history of ovarian cancer. Needs that have been identified in women with an abnormal TVS screening result deal with anticipation, emotional responses, role of the sonographer and impact of prior cancer experiences [50]. In examining social cognitive processing versus cognitive social health processing after an abnormal TVS screening, analyses found that greater distress was associated with greater social constraint [51]. Thus, psychological conditions that are apparently associated with ovarian screening are governed by different underlying factors in different women and not the screening result per se.

Future perspective

For the reader concluding that the ability to detect early ovarian cancers with TVS is very good, but has reservations about four to five screen-detected benign cases reaching surgery for every malignancy, the future rests on improvements to the PPV. Some of this improvement will come as confidence builds in making use of serial TVS with morphology indexing to allow for the resolution of ovarian structure interpreted as abnormal by TVS, and thus reduce the number of abnormal results going to surgery. Improvements may also arise as new biomarkers are used in conjunction with TVS to separate the identities of malignancies requiring surgery from those abnormalities that do not. Likewise, serial rises in serum biomarkers may be identified to be used as a prerequisite for surgery.

For readers concluding that the ability to detect early ovarian cancers with TVS is not very good, the future will depend on innovation and biodiscoveries of new ways to proceed.

For both of these groups, the changes that will accompany the Patient Protection and Affordable Care Act (PPACA) will be absolutely determinate for the future of ovarian cancer screening with TVS in the USA. Price and cost will likely be major drivers upon implementation of PPACA so that systems not only have the lowest unit costs, but also can be delivered with aggregate lowest cost will be the best candidates for acceptance. This is likely to mean that screening centers that are able to always achieve capacity screening will need to come into creation. Detection of ovarian tumors with TVS is very amenable to the screening center approach, as indicated by the Kentucky trial experience. Importantly, with the caregiver shortage that is anticipated when the PPACA brings access to at least a third more Americans, a screening center approach is likely to dove-tail well with new ways of providing care with efficiency of scale. A reasonable expectation is that results from follow-up in both the UKCTOCS and SCSOCS trials will demonstrate a survival benefit because both trials have demonstrated a shift to early stage disease. Application in Europe will fit well with a screening center approach through national health services. In light of much poorer ovarian cancer survivals in the UK relative to Canada and Australia [52], approaches that identify early stage disease with good prognosis will be capable of bringing needed improvement in ovarian cancer survival in these countries.

Executive summary

Screening

Screening is appropriate for disease that is grave and threatens well-being, and is curable if found early.

The screening method must be effective in finding disease that can be treated successfully.

Evaluating the effect of ovarian tumor detection on survival

Survival estimates need to be made only after sufficient time has passed for mortalities to have occurred.

Only cases that reflect the highest mortality risk should be included in survival analysis so that tumors with low malignant potentials are excluded.

Survival estimates should include both true positives and false negatives in the screened group.

The living versus dead status of all screening participants should be determined and cause of death established.

Kaplan–Meier survival analysis yields the most powerful method for comparing screened with unscreened groups.

Framework & findings: results from trials using ultrasound for the early detection of ovarian tumors

Similarities and differences exist in different trials with regard to population control versus randomized control and intent-to-treat versus per protocol analysis.

Three of the trials (Kentucky [70%], UKCTOCS [50%], SCSOCS [67%]) detected a significant shift to the discovery of early stage I and II disease, while one did not (PLCO [28%]).

High sensitivity ranging from 80–89% was achieved by these three trials, as well as high specificity >98%.

Positive predictive value was highest in the SCSOCS (47%) and Kentucky (20%) trials indicating, that one to four women without malignancy received surgery for each woman with a malignancy.

A majority of aggressive type II ovarian cancers were detected in each trial.

The Kentucky trial found a significant survival benefit, while the PLCO did not.

The PLCO trial

34,253 women were screened with recruiting at different times over an 8-year period to receive four annual screens using a TVS/CA-125 combination and two annual screens using CA-125 alone and received follow-up for a maximum of 13 years.

Not all women in the intervention group received all six rounds of screening.

In the usual care group, bimanual physical examination was stopped in year 6 out of the 14 years that were possible.

Not all screen-detected malignancies were treated.

PLCO survival analysis

Women in the screened group that developed ovarian cancer were considered in the survival analysis even if they were never screened or not treated because of how they were randomized.

No difference in mortality was observed between the screened and control groups.

Ovarian cancers that developed as long as 7 years or more after screening were counted as ovarian cancer deaths in the screening group.

The use of two final screens with CA-125 alone as a single trigger showed a much reduced ability to detect ovarian malignancies.

Questions worthy of discussion

Is the early detection of ovarian tumors using ultrasound effective?

– Three of the four trials reported the detection of early stage disease. Long-term follow-up from the Kentucky trial has demonstrated a survival benefit supporting effectiveness.

Are there differences in ultrasound screening algorithms that affect screening results?

– There were to be shorter delays to surgery in the Kentucky trial than in the PLCO trial so that progression to a later stage disease at diagnosis was less in the Kentucky trial. A single abnormal trigger from screening in the PLCO trial resulted in a lower positive predictive value than in the Kentucky trial that used serial TVS to advance a suspicious case to surgery.

Is the prevalence of ovarian cancer fatal to screening?

– Prevalence is really a cost question. Examination of the cost to screen the population utilized in the Kentucky trial shows that the savings gained by preventing progression of the early stage cases offsets 85–97% of the cost to screen the entire group. This offset is increased by additionally considering the dollar value of life lost so that the savings gained by screening more than pays for the cost of screening.

Does screening for ovarian tumors with ultrasound do more harm than good?

– Psychological conditions that are associated with ovarian screening are governed by underlying factors in different women and not the screening result per se.

Footnotes

Early detection of ovarian tumors using ultrasound

To obtain credit, you should first read the journal article. After reading the article, you should be able to answer the following, related, multiple-choice questions. To complete the questions (with a minimum 70% passing score) and earn continuing medical education (CME) credit, please go to www.medscape.org/journal/wh. Credit cannot be obtained for tests completed on paper, although you may use the worksheet below to keep a record of your answers. You must be a registered user on Medscape.org. If you are not registered on Medscape.org, please click on the New Users: Free Registration link on the left hand side of the website to register. Only one answer is correct for each question. Once you successfully answer all post-test questions you will be able to view and/or print your certificate. For questions regarding the content of this activity, contact the accredited provider, CME@medscape.net. For technical assistance, contact CME@webmd.net. American Medical Association's Physician's Recognition Award (AMA PRA) credits are accepted in the US as evidence of participation in CME activities. For further information on this award, please refer to http://www.ama-assn.org/ama/pub/category/2922.html. The AMA has determined that physicians not licensed in the US who participate in this CME activity are eligible for AMA PRA Category 1 Credit(s)™. Through agreements that the AMA has made with agencies in some countries, AMA PRA credit may be acceptable as evidence of participation in CME activities. If you are not licensed in the US, please complete the questions online, print the AMA PRA CME credit certificate and present it to your national medical association for review.

References

Papers of special note have been highlighted as:.

of interest.

of considerable interest

Pavlik

Saunders

Doran

The search for meaning – symptoms and transvaginal sonography screening for ovarian cancer. Cancer 115, 3689–3698 (2009).

Rossing

Wicklund

Cushing-Haugen

Weiss

. Predictive value of symptoms for early detection of ovarian cancer. J. Natl Cancer Inst. 102, 222–229 (2010).

Prorok

. Evaluation of screening program for the early detection of cancer. Natl Cancer Inst. Stat. Textbk Monog. 51, 267–328 (1984).

Hulka

. Cancer screening: degrees of proof and practical application. Cancer 62(Suppl.), 1776–1789 (1988).

Downey

Curtis

Lafferty

Herting

Engelberg

. The quality of dying and death (QODD) questionnaire. empirical domains and theoretical perspectives. J. Pain Symptom Manag. 39, 9–22 (2010).

Cooper

Nelson

Doran

Long-term survival and cost of treatment in patients with Stage IIIC epithelial ovarian cancer. Curr. Women's Health Rev. 5, 44–50 (2009).

10.

Siegel

Naishadham

Jemal

. Cancer statistics, 2012. CA Cancer J. Clin. 62, 10–29 (2012).

11.

Ryerson

Eheman

Burton

Symptoms, diagnoses, and time to key diagnostic procedures among older U.S. women with ovarian cancer. Obstet. Gynecol. 109, 1053–1061 (2007).

12.

Campbell

Bhan

Royston

Whitehead

Collins

. Transabdominal ultrasound screening for early ovarian cancer. Br. Med. J. 299, 1363–1367 (1989).

13.

Dellenbach

Nisand

Moreau

Transvaginal sonographically controlled ovarian follicle puncture for egg retrieval. Lancet 1, 1467–1471 (1984).

14.

Schwimer

Lebovic

. Transvaginal pelvic ultrasonography. J. Ultrasound Med. 3, 381–383 (1984).

15.

Schwimer

Lebovic

. TVS pelvic ultrasonography: accuracy in follicle and cyst size determination. J. Ultrasound Med. 4, 61–63 (1985).

16.

Jacobs

Bridges

Reynolds

Multimodal approach to screening for ovarian cancer. Lancet 331, 268–271 (1988).

17.

Higgins

van Nagell

Jr Donaldson

Transvaginal sonography as a screening method for ovarian cancer. Gynecol. Oncol. 34, 402–406 (1989).

18.

McCluggage

. The pathology of and controversial aspects of ovarian borderline tumors. Curr. Opin Oncol. 22, 462–472 (2010).

19.

Uzan

Kane

Rey

How to follow up advanced-stage borderline tumors? Mode of diagnosis of recurrence in a large series stage II–III serous borderline tumors of the ovary. Ann. Oncol. 22, 631–635 (2011).

20.

Kurman

Shih

. The origin and pathogenesis of epithelial ovarian cancer: a proposed unifying theory. Am. J. Surg. Pathol. 34, 433–443 (2010).

21.

Review of the origin of ovarian cancer.

22.

Landen

Birrer

Sood

. Early events in the pathogenesis of epithelial ovarian cancer. J. Clin. Oncol. 26, 995–1005 (2008).

23.

Review of the origin of ovarian cancer.

24.

Kikkawa

Nawa

Kajiyama

Shibata

Ino

Nomura

. Clinical characteristics and prognosis of mucinous tumors of the ovary. Gynecol. Oncol. 103, 171–175 (2006).

25.

Goodman

Shvetsov

. Incidence of ovarian, peritoneal and fallopian tube carcinomas in the United States, 1995–2004. Cancer Epidemiol. Biomarkers Prev. 18, 132–139 (2009).

26.

van Nagell

Miller

DeSimone

Long-term survival of women with epithelial ovarian cancer detected by ultrasonographic screening. Obstet. Gynecol. 118, 1212–1221 (2011).

27.

Major trial results: the Kentucky trial.

28.

Buys

Partridge

Black

Effect of screening on ovarian cancer mortality – the prostate, lung, colorectal and ovarian (PLCO) cancer screening randomized controlled trial. J. Am. Med. Assoc. 305, 2295–2303 (2011).

29.

Major trial results: the PLCO trial.

30.

Menon

Gentry-Maharaj

Hallett

Sensitivity and specificity of multimodal and ultrasound screening for ovarian cancer, and stage distribution of detected cancers: results of the prevalence screen of the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS). Lancet Oncol. 10, 327–340 (2009).

31.

Major trial results: the UKCTOCS trial.

32.

Kobayashi

Yamada

Sado

A randomized study of screening for ovarian cancer: a multicenter study in Japan. Int. J. Gynecol. Cancer 18, 414–420 (2008).

33.

Major trial results: the SCSOCS trial.

34.

Partridge

Kreimer

Greenlee

Results from four rounds of ovarian cancer screening in a randomized trial. Obstet. Gynecol. 113, 775–782 (2009).

35.

Partial early results from the PLCO trial.

36.

Menon

Gentry-Maharaj

Jacobs

. Ovarian cancer screening and mortality. J. Am. Med. Assoc. 306, 1544–1545 (2011).

37.

Printz

. CancerScope ASCO highlights new findings on cancer screening discoveries connected to ovarian, cervical and prostate cancer. Cancer 117, 3627–3629 (2011).

38.

Martzband

Groopman

. There is more to life than death. N. Engl. J. Med. 367, 987–989 (2012).

39.

Crum

Mckeon

Wian

. The oviduct and ovarian cancer: causality, clinical implications and “targeted prevention.” Clin. Obstet. Gynecol. 55, 24–35 (2012).

40.

Current review of the origin of ovarian cancer.

41.

Ueland

Depriest

Desimone

The accuracy of examination under anesthesia and transvaginal sonography in evaluating ovarian size. Gynecol. Oncol. 99, 400–403 (2005).

42.

Slomski

. Screening women for ovarian cancer still does more harm than good. J. Am. Med. Assoc. 307, 2474–2475 (2012).

43.

Humphrey

Hefland

Chan

BKS

Woolf

. Breast cancer screening: a summary of the evidence for the U.S. Preventive Services Task Force. Ann. Intern. Med. 137, 347–360 (2002).

44.

Elmore

Armstrong

Lehman

Fletcher

. Screening for breast cancer. JAMA 293, 1245–1256 (2005).

45.

Lehman

Gatsonis

Kuhl

MRI Evaluation of the contralateral breast in women with recently diagnosed breast cancer. N. Engl. J. Med. 356, 1295–1303 (2007).

46.

Bennett

Blanks

. Should a standard be defined for the positive predictive value (PPV) of recall in the UK NHS Breast Screening Programme? Breast 16, 55–59 (2007).

47.

Kopans

. The positive predictive value of mammography. Am. J. Roentgenol. 158, 521–526 (1992).

48.

Venkatesan

Philip

Kerlikowske

Sickles

Smith-Bindman

. Positive predictive value of specific mammographic findings according to reader and patient variables. Radiology 250, 648–657 (2009).

49.

Miller

Ware R

Smith

DeSimone

Performance of the American College of obstetricians and gynecologists' ovarian tumor referral guidelines with a multivariate index assay. Obstet. Gynecol. 117, 1298–1306 (2011).

50.

Results on clinical detection performance.

51.

Pavlik

van Nagell

DePriest

Participation in transvaginal ovarian cancer screening: compliance, correlation factors, and costs. Gynecol. Oncol. 57, 395–400 (1995).

52.

Pavlik

Johnson

DePriest

Andrykowski

Kryscio

van Nagell

. Continuing participation supports ultrasound screening for ovarian cancer. Ultrasound Obstet. Gynecol. 15, 354–364 (2000).

53.

Cohn

Kim

Resnick

O'Malley

Straughn

. At what cost does a potential survival advantage of bevacizumab make sense for the primary treatment of ovarian cancer? A cost–effectiveness analysis. J. Clin. Oncol. 29, 1247–1251 (2011).

54.

Yabroff

Bradley

Mariotto

Brown

Feuer

. Estimates and projections of value of life lost from cancer deaths in the United States. J. Natl Cancer Inst. 100, 1755–1762 (2008).

55.

Estimation of the value of life lost to ovarian cancer.

56.

Gentry-Maharaj

Sharma

Burnell

Volunteer acceptance of transvaginal scanning – United Kingdom Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) experience. Ultrasound Obstet. Gynecol. doi:10.1002/uog.12262 (2012) (Epub ahead of print).

57.

Pavlik

van Nagell

. Ovarian cancer screening – what women want. Int. J. Gynecol. Cancer 22(Suppl. 1), S21–S23 (2012).

58.

Dindo

Demartines

Clavien

. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann. Surg. 240, 205–213 (2004).

59.

Salsman

Pavlik

Boerner

Andrykowski

. Clinical, demographic, and psychological characteristics of new, asymptomatic partipants in a transvaginal ultrasound screening program for ovarian cancer. Prev. Med. 39, 315–322 (2004).

60.

Lykins

Pavlik

Andrykowski

. Validity of self-reports of return for routine repeat screening in an ovarian screening program. Cancer Epidemiol. Biomarkers Prev. 16, 490–493 (2007).

61.

Gaugler

Pavlik

Salsman

Andrykowski

. Pyschological and behavioral impact of receipt of a “normal” ovarian cancer screening test. Prev. Med. 42, 463–470 (2006).

62.

Andrykowski

Boerner

Salsman

Pavlik

. Psychological response to test results in an ovarian cancer screening program: a prospective, longitudinal study. Health Psychol. 23, 622–630 (2004).

63.

Ryan

Graves

Pavlik

Andrykowski

. Abnormal ovarian cancer screening test result: women's informational, psychological and practical needs. J. Psychosoc. Oncol. 25, 1–18 (2007).

64.

Andrykowski

Pavlik

. Response to an abnormal ovarian cancer-screening test result: test of the social cognitive processing and cognitive social health information processing models. Psychol. Health 26, 383–397 (2011).

65.

Coleman

Forman

Bryant

Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK, 1995–2007 (the International Cancer Benchmarking Partnership): an analysis of population-based cancer registry data. Lancet 377, 127–138 (2011).

66.

Sahr

. Consumer Price Index (CPI) Conversion Factors 1774 to estimated 2022 to Convert to Dollars of 2011. Oregon State University, OR, USA (2011). http://oregonstate.edu/cla/polisci/sites/default/files/faculty_files/sahr/cv2011.xls.

Early Detection of Ovarian Tumors Using Ultrasound

Abstract

Keywords

Screening

The disease significantly impacts public health

The disease has a high-to-intermediate probability of occurrence in the population

A critical point for disease detection must exist

Medical care is available if screening is positive

Patients are willing & able to undergo further evaluation & treatment

Ultrasonography

Factors for evaluating tumor detection performance

Evaluating the effect of ovarian tumor detection on survival

Framework & findings: results from trials using ultrasound for the early detection of ovarian tumors

The PLCO trial

PLCO survival analysis

Detection reported in the PLCO trial

Questions worthy of discussion

Is the early detection of ovarian tumors using ultrasound effective?

Are there differences in the application of an ultrasound screening algorithm that would affect screening results?

Is the prevalence of ovarian cancer fatal to screening?

Does screening for ovarian tumors with ultrasound do more harm than good?

Future perspective

Footnotes

Early detection of ovarian tumors using ultrasound

References