Background Overdiagnosis of clinically insignificant prostate cancer is considered a major potential drawback of prostate-specific antigen (PSA) screening. Quantitative estimates of the magnitude of this problem are, however, lacking. We estimated rates of prostate cancer overdiagnosis due to PSA testing that are consistent with the observed incidence of prostate cancer in the United States from 1988 through 1998. Overdiagnosis was defined as the detection of prostate cancer through PSA testing that otherwise would not have been diagnosed within the patient's lifetime. Methods We developed a computer simulation model of PSA testing and subsequent prostate cancer diagnosis and death from prostate cancer among a hypothetical cohort of two million men who were 60-84 years old in 1988. Given values for the expected lead time--that is, the time by which the test advanced diagnosis--and the expected incidence of prostate cancer in the absence of PSA testing, the model projected the increase in population incidence of prostate cancer associated with PSA testing. By comparing the model-projected incidence with the observed incidence derived from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) registry data, we determined the lead times and corresponding overdiagnosis rates that were consistent with the observed data. Results SEER data on prostate cancer incidence from 1988 through 1998 were consistent with overdiagnosis rates of approximately 29% for whites and 44% for blacks among men with prostate cancers detected by PSA screening. Conclusions Among men with prostate cancer that would be detected only at autopsy, these rates correspond to overdiagnosis rates of, at most, 15% in whites and 37% in blacks. The observed trends in prostate cancer incidence are consistent with considerable overdiagnosis among PSA-detected cases. However, the results suggest that the majority of screen-detected cancers diagnosed between 1988 and 1998 would have presented clinically and that only a minority of cases found at autopsy would have been detected by PSA testing.

/pdf/overdiagnosis-due-to-prostate-specific-antigen-screening-3yluhl4plh.pdf

Overdiagnosis Due to Prostate-Specific Antigen Screening: Lessons From U.S. Prostate Cancer Incidence Trends

Context The effect on mortality of screening for lung cancer with modern chest radiographs is unknown Objective To evaluate the effect on mortality of screening for lung cancer using radiographs in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial Design, Setting, and Participants Randomized controlled trial that involved 154 901 participants aged 55 through 74 years, 77 445 of whom were assigned to annual screenings and 77 456 to usual care at 1 of 10 screening centers across the United States between November 1993 and July 2001 The data from a subset of eligible participants for the National Lung Screening Trial (NLST), which compared chest radiograph with spiral computed tomographic (CT) screening, were analyzed Intervention Participants in the intervention group were offered annual posteroanterior view chest radiograph for 4 years Diagnostic follow-up of positive screening results was determined by participants and their health care practitioners Participants in the usual care group were offered no interventions and received their usual medical care All diagnosed cancers, deaths, and causes of death were ascertained through the earlier of 13 years of follow-up or until December 31, 2009 Main Outcome Measures Mortality from lung cancer Secondary outcomes included lung cancer incidence, complications associated with diagnostic procedures, and all-cause mortality Results Screening adherence was 866% at baseline and 79% to 84% at years 1 through 3; the rate of screening use in the usual care group was 11% Cumulative lung cancer incidence rates through 13 years of follow-up were 201 per 10 000 person-years in the intervention group and 192 per 10 000 person-years in the usual care group (rate ratio [RR]; 105, 95% CI, 098-112) A total of 1213 lung cancer deaths were observed in the intervention group compared with 1230 in usual care group through 13 years (mortality RR, 099; 95% CI, 087-122) Stage and histology were similar between the 2 groups The RR of mortality for the subset of participants eligible for the NLST, over the same 6-year follow-up period, was 094 (95% CI, 081-110) Conclusion Annual screening with chest radiograph did not reduce lung cancer mortality compared with usual care Trial Registration clinicaltrialsgov Identifier: NCT00002540

https://jamanetwork.com/journals/jama/articlepdf/1104579/joc15127_1865_1873.pdf

Screening by Chest Radiograph and Lung Cancer Mortality: The Prostate, Lung, Colorectal, and Ovarian (PLCO) Randomized Trial

Objective To determine the optimal method for quantifying and monitoring overdiagnosis in cancer screening over time.

Design Systematic review of primary research studies of any design that quantified overdiagnosis from screening for nine types of cancer. We used explicit criteria to critically appraise individual studies and assess strength of the body of evidence for each study design (double blinded review), and assessed the potential for each study design to accurately quantify and monitor overdiagnosis over time.

Data sources PubMed and Embase up to 28 February 2014; hand searching of systematic reviews.

Eligibility criteria for selecting studies English language studies of any design that quantified overdiagnosis for any of nine common cancers (prostate, breast, lung, colorectal, melanoma, bladder, renal, thyroid, and uterine); excluded case series, case reports, and reviews that only reported results of other studies.

Results 52 studies met the inclusion criteria. We grouped studies into four methodological categories: (1) follow-up of a well designed randomized controlled trial (n=3), which has low risk of bias but may not be generalizable and is not suitable for monitoring; (2) pathological or imaging studies (n=8), drawing conclusions about overdiagnosis by examining biological characteristics of cancers, a simple design limited by the uncertain assumption that the measured characteristics are highly correlated with disease progression; (3) modeling studies (n=21), which can be done in a shorter time frame but require complex mathematical equations simulating the natural course of screen detected cancer, the fundamental unknown question; and (4) ecological and cohort studies (n=20), which are suitable for monitoring over time but are limited by a lack of agreed standards, by variable data quality, by inadequate follow-up time, and by the potential for population level confounders. Some ecological and cohort studies, however, have addressed these potential weaknesses in reasonable ways.

Conclusions Well conducted ecological and cohort studies in multiple settings are the most appropriate approach for quantifying and monitoring overdiagnosis in cancer screening programs. To support this work, we need internationally agreed standards for ecological and cohort studies and a multinational team of unbiased researchers to perform ongoing analysis.

https://www.bmj.com/content/bmj/350/bmj.g7773.full.pdf

Quantifying and monitoring overdiagnosis in cancer screening: a systematic review of methods

The introduction of the prostate-specific antigen (PSA) test has led to dramatic changes in the incidence of prostate cancer in the United States. In this article, we use information on the increase and subsequent decline in prostate cancer incidence following the adoption of PSA to estimate the lead time associated with PSA screening. The lead time is a key determinant of the likelihood of overdiagnosis, one of the main costs associated with the PSA test. Our approach conceptualizes observed incidence as the sum of the secular trend in incidence, which reflects incidence in the absence of PSA, and the excess incidence over and above the secular trend, which is a function of population screening patterns and the unknown lead time. We develop a likelihood model for the excess incidence given the secular trend and use it to estimate the mean lead time under specified distributional assumptions. We also develop a likelihood model for observed incidence and use it to simultaneously estimate the mean lead time together with a smooth secular trend. Variances and confidence intervals are estimated via a parametric bootstrap. Our results indicate an average lead time of approximately 4.59 years (95% confidence interval [3.24, 5.93]) for whites and 6.78 years [5.42, 8.20] for blacks with a corresponding secular trend estimate that is fairly flat after the introduction of PSA screening. These estimates correspond to overdiagnosis frequencies of approximately 22.7% and 34.4% for screen-detected whites and blacks, respectively. Our results provide the first glimpse of a plausible secular trend in prostate cancer incidence and suggest that, in the absence of PSA screening, disease incidence would not have continued its historic increase, rather it would have leveled off in accordance with changes in prostate patterns of care unrelated to PSA.

Estimating lead time and overdiagnosis associated with PSA screening from prostate cancer incidence trends.

Knowledge of the likelihood that a screening-detected case of cancer has been overdiagnosed is vitally important to make treatment decisions and develop screening policy. An overdiagnosed case is an excess case detected by screening. Estimates of the frequency of overdiagnosis in breast and prostate cancer screening vary greatly across studies. This article identifies features of overdiagnosis studies that influence results and shows their effect by using published research. First, different ways to define and measure overdiagnosis are considered. Second, contextual features and how they affect overdiagnosis estimates are examined. Third, the effect of estimation approach is discussed. Many studies use excess incidence under screening as a proxy for overdiagnosis. Others use statistical models to make inferences about lead time or natural history and then derive the corresponding fraction of cases that are overdiagnosed. This article concludes with questions that readers of overdiagnosis studies can use to evaluate the validity and relevance of published estimates and recommends that authors of studies quantifying overdiagnosis provide information about these features.

/pdf/influence-of-study-features-and-methods-on-overdiagnosis-1r17dxr0b6.pdf

Influence of Study Features and Methods on Overdiagnosis Estimates in Breast and Prostate Cancer Screening

The model that specifies that cancer incidence, I, is the convolution of the preclinical incidence, g, and the density of time in the preclinical phase, f, has frequently been utilized to model data from cancer screening trials and to estimate such quantities as sojourn time, lead time, and sensitivity. When this model is fit to the above data, the parameters of f as well as the parameter(s) governing screening sensitivity must be estimated. Previously, g was either assumed to be equal to clinical incidence or assumed to be a constant or exponential function that also had to be estimated. Here we assume that the underlying incidence, I, in the study population (in the absence of screening) is known. With I known, g then becomes a function of f, which can be solved for using (numerical) deconvolution, thus eliminating the need to estimate g or make assumptions about it. Since numerical deconvolution procedures may be highly unstable, however, we incorporate a smoothing procedure that produces a realistic g function while still closely reproducing the original incidence function I upon convolution with f. We have also added the concept of competing mortality to the convolution model. This, along with the realistic preclinical incidence function described above, results in more accurate estimates of sojourn time and lead time and allows for estimation of quantities related to overdiagnosis, which we define here.

Estimation and Prediction for Cancer Screening Models Using Deconvolution and Smoothing

The standard convolution model of disease natural history posits an asymptomatic (preclinical) and a symptomatic (clinical) state. An augmented model includes, in both the preclinical and clinical states, an early and late stage of disease. In the case of cancer, the early stage would generally correspond to the organ-confined stages before there is evidence of cancer spread. We compute the number of screen-detected (preclinical) and clinical cases in the early and late stages expected under a given screening program and show how the model can be fit to data from a screening trial using maximum likelihood. We also develop expressions for sojourn time, lead time, and overdiagnosis in the context of the model, where each of the above concepts incorporates disease stage. As an example, we fit the model to data from the Mayo Lung Cancer Screening trial.

Paul F. Pinsky

Papers

Estimation and Prediction for Cancer Screening Models Using Deconvolution and Smoothing

An early- and late-stage convolution model for disease natural history.