Deriving benefit of early detection from biomarker-based prognostic models.

Many prognostic models for cancer use biomarkers that have utility in early detection. For example, in prostate cancer, models predicting disease-specific survival use serum prostate-specific antigen levels. These models typically show that higher marker levels are associated with poorer prognosis. Consequently, they are often interpreted as indicating that detecting disease at a lower threshold of the biomarker is likely to generate a survival benefit. However, lowering the threshold of the biomarker is tantamount to early detection. For survival benefit to not be simply an artifact of starting the survival clock earlier, we must account for the lead time of early detection. It is not known whether the existing prognostic models imply a survival benefit under early detection once lead time has been accounted for. In this article, we investigate survival benefit implied by prognostic models where the predictor(s) of disease-specific survival are age and/or biomarker level at disease detection. We show that the benefit depends on the rate of biomarker change, the lead time, and the biomarker level at the original date of diagnosis as well as on the parameters of the prognostic model. Even if the prognostic model indicates that lowering the threshold of the biomarker is associated with longer disease-specific survival, this does not necessarily imply that early detection will confer an extension of life expectancy.

Cost-effectiveness of microsatellite instability screening as a method for detecting hereditary nonpolyposis colorectal cancer.

BACKGROUND: The National Cancer Institute has published consensus guidelines for universal screening for hereditary nonpolyposis colorectal cancer (HNPCC) in patients with newly diagnosed colorectal cancer. OBJECTIVE: To determine the cost-effectiveness of screening compared with standard care in eligible patients with colorectal cancer and their siblings and children. DESIGN: Cost-effectiveness analysis. DATA SOURCES: National colorectal cancer registry data, the Creighton International Hereditary Colorectal Cancer Registry, Medicare claims records, and published literature. TARGET POPULATION: Patients with newly diagnosed colorectal cancer and their siblings and children. TIME HORIZON: Lifetime (varies depending on age at screening). PERSPECTIVE: Societal. INTERVENTIONS: Initial office-based screening to determine eligibility (based on personal and family cancer history), followed by tumor testing for microsatellite instability. Those with microsatellite instability were offered genetic testing for HNPCC. Siblings and children of patients with cancer and the HNPCC mutation were offered genetic testing, and those who were found to carry the mutation received lifelong colorectal cancer screening. MEASUREMENTS: Life-years gained. RESULTS OF BASE-CASE ANALYSIS: When only the patients with cancer were considered, cost-effectiveness of screening was $42 210 per life-year gained. When patients with cancer and their siblings and children were considered together, cost-effectiveness increased to $7556 per life-year gained. RESULTS OF SENSITIVITY ANALYSIS: The model was most sensitive to the estimated survival gain from screening siblings and children, to the prevalence of HNPCC mutations among patients with newly diagnosed cancer, and to the discount rate. In probabilistic analysis, the 90% CI for the cost-effectiveness of screening patients with cancer plus their relatives was $4874 to $21 576 per life-year gained. CONCLUSION: Screening patients with newly diagnosed colorectal cancer for HNPCC is cost-effective, especially if the benefits to their immediate relatives are considered.

Serial prostate specific antigen, free-to-total prostate specific antigen ratio and complexed prostate specific antigen for the diagnosis of prostate cancer.

PURPOSE: The free-to-total prostate specific antigen (PSA) ratio and complexed PSA have been introduced as adjuncts to total PSA for prostate cancer screening. Little data exist on the use of these tests for serial PSA screening. We compared serial total PSA, the free-to-total PSA ratio and calculated complexed PSA in men diagnosed with prostate cancer and matched controls in a population based study. MATERIALS AND METHODS: We identified 90 men diagnosed with prostate cancer between 1988 and 1996 with at least 3 serial serum samples obtained at 2-year intervals who were participants in the beta-Carotene and Retinol Efficacy Trial for the prevention of lung cancer. Samples were available up to 10 years before diagnosis. A total of 90 age matched men from the same cohort without prostate carcinoma were identified as controls. Free and total PSA was measured by the Abbott AxSYM system. RESULTS: Baseline demographics of cases and controls were similar. At baseline and diagnosis the men with prostate cancer had higher total and complexed PSA, and a lower free-to-total PSA ratio than controls. Mean followup was 5.2 years in cases and 5.5 in controls. The yearly change in PSA parameters in cases versus controls was 20.7% versus 3.5% for total, -3.4% versus 0.2% for free-to-total and 21.5% versus 3.4% for complexed PSA (p <0.0001). At diagnosis PSA alone was estimated to perform with more than 90% specificity in our model. CONCLUSIONS: In this population based study total PSA was superior to the free-to-total PSA ratio for predicting the development of prostate cancer. While serial changes in free-to-total PSA ratios with time were statistically significantly different in men diagnosed with prostate cancer and controls, the magnitude of these serial changes were slight enough to render them clinically insignificant.

The impact of including future medical care costs when estimating the costs attributable to a disease: a colorectal cancer case study.

A source of controversy in the economic literature concerns whether to include or exclude future medical care costs when computing attributable costs for lifesaving interventions. Although it is hypothesized that including future medical care costs will offset the cost savings achieved through prevention, the magnitude of the effect is not known. The objectives of the present study are to develop a methodology for estimating the excess costs of care among colorectal cancer patients, including and excluding future costs of care, and comparing these results with previous studies that do not include costs in added years of life. Subjects in the study included those identified with colorectal cancer drawn from the Surveillance, Epidemiology and End Results (SEER)-Medicare database and an age- and gender-matched control group drawn from the general Medicare population. Using the Kaplan-Meier Sample Average estimator, we directly estimate expected 11-year costs, and then, with the addition of some simple extrapolating assumptions, determine expected 25-year costs. The latter time horizon captures lifetime costs for over 90% of the cohort. Males results for discounted, stage-specific 11- versus 25-year excess costs: in situ, 22411 dollars versus 23494 dollars; Stage 1, 29365 dollars versus 32510 dollars; Stage 2, 28114 dollars versus 25263 dollars; Stage 3, 27397 dollars versus 19647 dollars; Stage 4, 3006 dollars versus 7837 dollars. Trends were similar for females. It can be concluded that adding costs of care in future years for those whose colorectal cancer is prevented owing to screening greatly alters the estimate of lifetime excess costs for colorectal cancer patients, and can produce negative results for advanced stage disease. The results emphasize the need to adopt a standard approach for dealing with future costs when evaluating lifesaving interventions for cost-effectiveness analyses.

Simulation modeling of outcomes and cost effectiveness.

Modeling will continue to be used to address important issues in clinical practice and health policy issues that have not been adequately studied with high-quality clinical trials. The apparent ad hoc nature of models belies the methodologic rigor that is applied to create the best models in cancer prevention and care. Models have progressed from simple decision trees to extremely complex microsimulation analyses, yet all are built using a logical process based on objective evaluation of the path between intervention and outcome. The best modelers take great care to justify both the structure and content of the model and then test their assumptions using a comprehensive process of sensitivity analysis and model validation. Like clinical trials, models sometimes produce results that are later found to be invalid as other data become available. When weighing the value of models in health care decision making, it is reasonable to consider the alternatives. In the absence of data, clinical policy decisions are often based on the recommendations of expert opinion panels or on poorly defined notions of the standard of care or medical necessity. Because such decision making rarely entails the rigorous process of data collection, synthesis, and testing that is the core of well-conducted modeling, it is usually not possible for external audiences to examine the assumptions and data that were used to derive the decisions. One of the modeler's most challenging tasks is to make the structure and content of the model transparent to the intended audience. The purpose of this article is to clarify the process of modeling, so that readers of models are more knowledgeable about their uses, strengths, and limitations.

Quality of life in survivors of colorectal carcinoma.

BACKGROUND: Colon carcinoma is a common malignancy that accounts for a substantial share of all cancer-related morbidity and mortality. However, little is known with regard to general and disease specific quality of life in survivors of colorectal carcinoma, particularly from community-based samples of cases across stage and survival times from diagnosis. METHODS: Subjects with colorectal carcinoma were recruited from the National Cancer Institute's Surveillance, Epidemiology, and End Results cancer registry. Subjects completed two self-administered surveys: the Functional Assessment of Cancer Therapy Scales for Colorectal Cancer (FACT-C) and the Health Utilities Index (HUI) Mark III. RESULTS: One hundred seventy-three respondents (average age: 70.4 years, 71.4% female) completed the survey. In the first 3 years after diagnosis, quality of life was lower and varied substantially among respondents. After 3 years, respondents in all TNM stages of disease except Stage IV reported a relatively uniform and high quality of life. Pain, functional well-being, and social well-being were affected most substantially across all stages and times from diagnosis. Low income status was associated with worse outcomes for pain, ambulation, and social and emotional well-being. Only emotional well-being scores improved significantly over time in both surveys. CONCLUSIONS: Those individuals who achieve a long term remission from colorectal carcinoma may experience a relatively high quality of life, although deficits remain for several areas, particularly in those of low socioeconomic status. Sampling design may have excluded the most severely ill patients.

The effectiveness of mammography promotion by volunteers in rural communities.

INTRODUCTION: The Community Trial of Mammography Promotion assessed the effectiveness of mammography promotion by community volunteer groups in rural areas. Three interventions were tested. One used an individual counseling strategy, one used a community activities strategy, and a third combined the two strategies. METHODS: The effects of the interventions were tested by randomizing 40 communities either to the study interventions or to a control group. A cohort of 352 women from each community was randomly selected and used to evaluate the interventions' effectiveness. Of these, 6592 women were eligible for screening mammography at baseline and follow-up and were successfully interviewed prior to and after study intervention activities. RESULTS: Although the interventions did not significantly increase women's overall use of mammography, the community activities intervention increased use at follow-up by regular users over baseline by 2.9% (p = 0.01). Intervention appears to have increased the use of mammography among certain groups of women who were not regular users at baseline, including those in communities without female physicians (10% to 16%; p < 0.05), and among women with no health insurance (10% to 23%; p

Reduced dose intravenous immunoglobulin does not decrease transplant-related complications in adults given related donor marrow allografts.

Graft-vs.-host disease (GVHD) and infection are major complications of allogeneic bone marrow transplantation. Intravenous immunoglobulin (IVIg) given at a dose of 500 mg/kg/wk has been shown to decrease the risk of acute GVHD, interstitial pneumonia, and infection in adults early after allogeneic transplantation. The current study is a controlled trial to determine whether a lower total dose of IVIg given with pretransplant loading reduces the incidence of transplant-related complications. In a randomized trial of 241 patients > or =20 years of age who were given related donor marrow allografts, 121 individuals receiving Ig prophylaxis (500 mg/kg/d loading from day -6 to -1 and then 100 mg/kg every 3 days from day 3 to 90) were compared with 120 control patients who did not receive IVIg. Randomization was stratified by human leucocyte antigen-matching, remission status of malignancy, GVHD prophylaxis, and cytomegalovirus (CMV) serology. The study was powered to detect a reduction in acute GVHD by 18% and a decrease in transplant-related mortality by 17%. Pretransplant IVIg loading and posttransplant maintenance achieved median serum IgG levels >1350 mg/dL, which were approximately twofold greater than the untreated controls (p<0.01). White blood cell and platelet recoveries were similar for the two groups, although control patients required fewer units of platelets per day (2.5 vs. 3.3, p = 0.008). No significant differences in the incidence of CMV infection, interstitial pneumonia, or bacteremia were observed. The incidence of acute GVHD did not differ between the two groups; however, acute GVHD was less frequent among IVIg recipients achieving maximum serum IgG levels >3000 mg/dL (60 vs. 79%). Neither transplant-related mortality nor disease-free survival was significantly altered by Ig prophylaxis. However, the cumulative incidence of relapse of malignancy was higher in IVIg recipients than in controls (31 vs. 18%, p = 0.03). Multivariable regression analysis demonstrated a 1.89 increased relative risk of relapse for individuals given IVIg (p = 0.021). We conclude that pretransplant loading and a shorter course and lower total dose of IVIg prophylaxis did not appear to decrease the risk of acute GVHD or mortality among adults receiving related donor marrow transplants. Note, IVIg administration may be associated with an increased risk of recurrent malignancy, a finding that warrants further investigation.

Obtaining long-term disease specific costs of care: application to Medicare enrollees diagnosed with colorectal cancer.

OBJECTIVES: This study develops estimates of long-term, cancer-related treatment cost using a modeling approach and data from the SEER-Medicare linked database. The method is demonstrated for colorectal cancer. METHODS: Data on Medicare payments were obtained for colorectal cancer patients for the years 1990 to 1994 from the SEER-Medicare linked database. Claims payment data for control subjects were obtained for Medicare enrollees without cancer residing in the same areas as patients. Estimates of long-term cost (< or = 25 years following the date of diagnosis) were obtained by combining treatment/phase-specific cost estimates with estimates of long-term survival from SEER. Treatment phases were defined as initial care, terminal care, and continuing care. Cancer-related estimates for each phase were obtained by subtracting costs for control subjects from the observed costs for cancer patients, matching on age group, gender, and registry area. Estimates of long-term cost < or = 11 years obtained by this method were compared with 11-year estimates obtained by application of the Kaplan-Meier sample average (KMSA) method. RESULTS: The mean initial-phase cancer-related cost was approximately $18,000 but was higher among patients with more advanced cancer. The mean continuing-phase cancer-related cost was $1,500 per year and declined with increasing age, but was higher on an annual basis among persons with later stages of cancer and shorter survival time. The mean terminal-phase cancer-related cost was $15,000 and declined with both age at death and more advanced stage at diagnosis. After the phase-specific estimates were combined, the average long-term cancer-related cost was $33,700 ($31,300 at 3% discount rate) for colon cancer compared with $36,500 ($33,800 at 3% discount rate) for cancer of the rectum. This represented about half of the total long-term cost for Medicare enrollees diagnosed with this disease. Long-term cost was highest for Stage III cancer and lowest for in situ cancer. Eleven-year cancer-related costs estimated by the KMSA method were similar to estimates using the phase-based approach. CONCLUSIONS: This paper demonstrates that valid estimates of cancer-related long-term cost can be obtained from administrative claims data linked to incidence cancer registry data.

On the use of survival analysis techniques to estimate medical care costs.

Measurement of treatment costs is important in the evaluation of medical interventions. Accurate cost estimation is problematic, when cost records are incomplete. Methods from the survival analysis literature have been proposed for estimating costs using available data. In this article, we clarify assumptions necessary for validity of these techniques. We demonstrate how assumptions needed for valid survival analysis may be violated when these methods are applied to cost estimation. Our observations are confirmed through simulations and empirical data analysis. We conclude that survival analysis approaches are not generally appropriate for the analysis of medical costs and review several valid alternatives.

Community attitudes and mammography use: does it really matter what other people think?

This study examined whether community characteristics, particularly community attitudes regarding mammography use, are associated with women's use of mammography in rural communities. Forty communities in predominantly rural areas of Washington State were selected for inclusion in this study based on their size and distance from an urban center. Characteristics of the communities were assessed as were characteristics of women living in the communities. From each community, random samples of 352 women between 50 and 80 years old participated by completing a telephone survey that included questions on a variety of topics, including their use of mammography. Logistic regression analyses revealed community of residence to be a significant predictor of individual women's mammography use after adjusting for individual level predictors of mammography use including age, education, employment, marital status, financial situation, and ease of access to medical services. An examination of the influence of community characteristics revealed women living in communities with supportive community attitudes towards mammography use report higher levels of mammography use than do women living in communities with less supportive attitudes. The presence or absence of male or female physicians or of mammography facilities in a local community was not associated with statistically significant effects on women's mammography use. Community attitudes are associated with mammography use in rural communities. Public health interventions that change community attitudes may have effects that extend beyond the people directly contacted by these interventions.

Serial prostate specific antigen screening for prostate cancer: a computer model evaluates competing strategies.

PURPOSE: We compare prostate specific antigen (PSA) screening strategies in terms of expected years of life saved with screening, number of screens, number of false-positive screens and rates of over diagnosis, defined as detection by PSA screening of patients who would never have been diagnosed without screening. MATERIALS AND METHODS: A computer model of disease progression, clinical diagnosis, PSA growth and PSA screening was used. Under baseline conditions, when screening is not considered, the model replicates clinical diagnosis and disease mortality rates recorded by the Surveillance, Epidemiology and End Results Program of the National Cancer Institute in the mid 1980s. RESULTS: Biannual screening with PSA greater than 4.0 ng./ml. was projected to reduce the number of screens and false-positive tests by almost 50% relative to annual screening while retaining 93% of years of life saved. With annual screening use of an age specific bound for PSA to consider a test positive instead of the standard 4.0 ng./ml. was projected to reduce false-positive screens by 27% and over diagnosis by a third while retaining almost 95% of years of life saved. Sensitivity analyses did not change the relative efficacy of biannual screening. CONCLUSIONS: Under the model assumptions biannual PSA screening is a cost-effective alternative to annual PSA screening for prostate cancer. With annual screening use of an age specific bound for PSA positivity appears to reduce false-positive results and over diagnosis rates sharply relative to a bound of 4 ng./ml. while retaining most of the survival benefits.

Incorporating the time dimension in receiver operating characteristic curves: a case study of prostate cancer.

Early diagnosis of disease has potential to reduce morbidity and mortality. Biomarkers may be useful for detecting disease at early stages before it becomes clinically apparent. Prostate-specific antigen (PSA) is one such marker for prostate cancer. This article is concerned with modeling receiver operating characteristic (ROC) curves associated with biomarkers at various times prior to the time at which the disease is detected clinically, by two methods. The first models the biomarkers statistically using mixed-effects regression models, and uses parameter estimates from these models to estimate the time-specific ROC curves. The second directly models the ROC curves as a function of time prior to diagnosis and may be implemented using software packages with binary regression or generalized linear model routines. The approaches are applied to data from 71 prostate cancer cases and 71 controls who participated in a lung cancer prevention trial. Two biomarkers for prostate cancer were considered: total serum PSA and the ratio of free to total PSA. Not surprisingly, both markers performed better as the interval between PSA measurement and clinical diagnosis decreased. Although the two markers performed similarly eight years prior to diagnosis, it appears that total PSA performed better than the ratio measure at times closer to diagnosis. The area under the ROC curve was consistently greater for total PSA than for the ratio four and two years prior to diagnosis and at the time of diagnosis.

Cancer surveillance series: interpreting trends in prostate cancer--part III: Quantifying the link between population prostate-specific antigen testing and recent declines in prostate cancer mortality.

BACKGROUND: The objective of this study was to investigate the circumstances under which dissemination of prostate-specific antigen (PSA) testing, beginning in 1988, could plausibly explain the declines in prostate cancer mortality observed from 1992 through 1994. METHODS: We developed a computer simulation model by use of information on population-based PSA testing patterns, cancer detection rates, average lead time (the time by which diagnosis is advanced by screening), and projected decreased risk of death associated with early diagnosis of prostate cancer through PSA testing. The model provides estimates of the number of deaths prevented by PSA testing for the 7-year period from 1988 through 1994 and projects what prostate cancer mortality for these years would have been in the absence of PSA testing. RESULTS: Results were generated by assuming a level of screening efficacy similar to that hypothesized for the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial. Under this assumption, the projected mortality in the absence of PSA testing continued the increasing trend observed before 1991 only when it was assumed that the mean lead time was 3 years or less. Projected mortality trends in the absence of PSA screening were not consistent with pre-1991 increasing trends for lead times of 5 years and 7 years. CONCLUSIONS: When screening is assumed to be at least as efficacious as hypothesized in the PLCO trial, it is unlikely that the entire decline in prostate cancer mortality can be explained by PSA testing based on current beliefs concerning lead time. Only very short lead times would produce a decline in mortality of the magnitude that has been observed.

Assessing uncertainty in microsimulation modelling with application to cancer screening interventions.

Microsimulation is fast becoming the approach of choice for modelling and analysing complex processes in the absence of mathematical tractability. While this approach has been developed and promoted in engineering contexts for some time, it has more recently found a place in the mainstream of the study of chronic disease interventions such as cancer screening. The construction of a simulation model requires the specification of a model structure and sets of parameter values, both of which may have a considerable amount of uncertainty associated with them. This uncertainty is rarely quantified when reporting micro-simulation results. We suggest a Bayesian approach and assume a parametric probability distribution to mathematically express the uncertainty related to model parameters. First, we design a simulation experiment to achieve good coverage of the parameter space. Second, we model a response surface for the outcome of interest as a function of the model parameters using the simulation results. Third, we summarize the variability in the outcome of interest, including variation due to parameter uncertainty, using the response surface in combination with parameter probability distributions. We illustrate the proposed method with an application of a microsimulator designed to investigate the effect of prostate specific antigen (PSA) screening on prostate cancer mortality rates.

Asymptomatic incidence and duration of prostate cancer.

Prostate cancer is known as a disease with an extremely high prevalence relative to its clinical incidence in the population. The combination of preclinical incidence and duration that could yield this phenomenon is of tremendous interest to researchers trying to understand the natural history of the disease and to develop efficient screening strategies. In this article, the authors present estimates of the age-specific asymptomatic incidence and average preclinical duration of prostate cancer. The methodological approach is to first estimate the age-specific incidence of new (stage AI) prostate cancers using preclinical prevalence data from autopsy studies performed between 1941 and 1964 and clinical incidence data for the years 1960-1986 from the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute. Then, the preclinical prevalence estimates are divided by the derived preclinical incidence estimates to yield estimates of the average duration of asymptomatic disease. The estimated mean duration among white men is between 11 and 12 years and appears to be approximately 1 year shorter for blacks than for whites. Comparison of the lifetime risks of preclinical and clinical disease suggests that approximately 75% of prostate cancers will never become diagnosed if clinical incidence remains at levels observed in 1984-1986, prior to the introduction of prostate-specific antigen (PSA) screening in the population.

Analysis of case-control studies of screening: impact of misspecifying the duration of detectable preclinical pathologic changes.

In case-control studies of screening to prevent cancer mortality, exposure is ideally defined as screening that takes place within that period prior to diagnosis during which the cancer is potentially detectable using the screening modality under study. This interval has been called the detectable preclinical period (DPP). Misspecifying the duration of the DPP can bias the results of such studies. This article quantifies the impact of incorrectly estimating the duration of the DPP or using the correct average DPP but failing to consider its variability. The authors developed a computer simulation model of disease incidence and mortality with and without screening. The authors then selected cases and controls from the generated population and compared their screening histories. The results indicate that underestimation of the duration of the DPP generally leads to greater bias than does overestimation, but in both instances the extent of the bias is modified by the relative length of the DPP and the average interscreening interval. In practice, the authors recommend that to prevent a falsely low estimate of the effectiveness of a screening test in reducing mortality, a high percentile of the DPP distribution be used when analyzing the results of case-control studies of screening.

Optimizing sampling strategies for estimating quality-adjusted life years.

Accurate estimation of quality of life is critical to cost-effectiveness analysis. Nevertheless, development of sampling algorithms to maximize the accuracy and efficiency of estimated quality of life has received little consideration to date. This paper presents a method to optimize sampling strategies for estimating quality-adjusted life years. In particular, the authors address the questions of when to sample and how many observations to sample at each sampling time, assuming realistically that the sample variance of quality of life is not constant over time. The method is particularly useful for the design problems researchers face when time or research budget constraints limit the number of individuals that can be surveyed to estimate quality of life. The article focuses on cross-sectional sampling. The method proposed requires some knowledge of survival in the population of interest, the approximate variances in utilities at various points along the curve, and the general shape of the quality-adjusted survival curve. Such data are frequently available from disease registries, the literature, or previous studies.