Combining CA 125 and SMR serum markers for diagnosis and early detection of ovarian carcinoma.

OBJECTIVES: The serum tumor marker CA 125 is elevated in most clinically advanced ovarian carcinomas. Because these elevations may precede clinical detection by a year or more, CA 125 is potentially useful for early detection as part of an ovarian cancer screening program. However, CA 125 is often not elevated in clinically detected cancer and is frequently elevated in women with benign ovarian tumors. CA 125 may be more useful in conjunction with one or more other tumor biomarkers. Additional markers could play a role if, when used with CA 125, they identify some carcinomas missed by CA 125 (i.e., they improve sensitivity), rule out false positives (i.e., improve specificity), or are able to detect the same cancers earlier. METHODS: We have evaluated a composite marker (CM) that combines CA 125 and a previously described soluble mesothelin related (SMR) marker in sera from 52 ovarian cancer cases, 43 controls with benign ovarian tumors, and 220 normal risk controls who participated in a screening program, including 25 healthy women having two serum samples collected 1 year apart. CA 125, SMR, and CM were evaluated for their ability to identify clinical disease and for their temporal stability, which assesses their ability to obtain even greater sensitivity when used in a longitudinal screening program. RESULTS: CM has the best sensitivity, with specificity equal to CA 125. Importantly, CM has temporal stability at least as high as CA 125. CONCLUSION: The CM may outperform CA 125 alone in a longitudinal screening program as well as in a diagnostic setting.

Ovarian cancer tumor marker behavior in asymptomatic healthy women: implications for screening.

Ovarian cancer screening protocols generally have been limited by inadequate recognition of the normal behavior of tumor markers in women at risk of ovarian cancer. We have characterized the behavior of five serum tumor markers in a large cohort of healthy women and examined the implications for screening. Serial measurements of CA125, HER-2/neu, urinary gonadotropin peptide, lipid-associated sialic acid, and Dianon marker 70/K were obtained during 6 years of follow-up of 1257 healthy women at high risk of ovarian cancer. We analyzed individual-specific tumor marker behavior and explored methods that can exploit this information to develop individual-specific screening rules. These five tumor markers behaved approximately independently. Substantial heterogeneity was observed among women in the behavior of each tumor marker, particularly CA125. Intraclass correlation (ICC), or the proportion of total variability that occurs between individuals, was approximately 0.6 for log-transformed CA125 and urinary gonadotropin peptide, and less than 0.4 for the other markers. This degree of tumor marker heterogeneity among healthy women, and the relative independence of these markers, has important implications for screening and diagnostic tests. Independence of markers results in the clinically useful fact that the combined false positive rate from screening with multiple markers may be estimated by the sum of individual false positive rates. Heterogeneity of tumor marker patterns in healthy women implies that a fixed screening cutoff level is suboptimal to a degree that depends strongly on ICC. Using information on longitudinal measurements and ICC, individual-specific screening rules may be developed with the potential to improve early detection of ovarian cancer.

Instrumental variables when evaluating screening trials: estimating the benefit of detecting cancer by screening.

When evaluating the benefit of detecting cancer by screening we try to answer the question, 'what would a screen detected subject's outcome have been if his/her cancer had progressed to clinical detection'. By 'outcome' we mean survival time, cancer size and stage, lead time effects and more. Because only an unethical study can answer it directly, researchers have attempted to answer the question indirectly using data from randomized cancer screening studies (subjects randomized to study (screened) or control (not screened)). Inferences are made by first selecting the cancer cohort (those subjects who are found to have cancer), then comparing subjects having screen detected cancers to subjects having clinically detected cancers. However, there are two difficulties with this approach: (i) because screening (intends to) detect cancers early, at the trial's end the study group contains more cancer cases than the control group and so the cancer cohort has some unidentified control subjects missing (that is, subjects having cancer during the screening period that have not yet been clinically detected); (ii) because screen detected cancers (may) differ from clinically detected cancers, the comparison group should include only a (non-identified) subset of the cancer cohort's control subjects (that is, only those control subjects having cancers that would have been screen detected). Statistical literature acknowledges these difficulties and attempts to solve them separately, but without success; those methods do not yield meaningful causal inferences and admit substantial bias. Recently, Angrist, Imbens and Rubin and Imbens and Rubin provide a framework for instrumental variable methods that we interpret as allowing us to make causal inferences with incompletely identified comparison groups. We apply their framework to evaluating cancer screening trials and find that we may simultaneously accommodate both difficulties while giving a meaningful answer to the question posed above. Using data from a breast cancer screening trial we demonstrate the general method with a variety of outcome measures and extensions.

Prostate-specific antigen for prostate cancer screening. Do physician characteristics affect its use?

PURPOSE: Screening for prostate cancer with the prostate-specific antigen (PSA) test remains controversial. This controversy is reflected in a lack of consensus in the medical literature and among professional and policy organizations regarding routine screening by PSA. It is not known how physicians respond when recommendations from experts are inconsistent. METHODS: A questionnaire was mailed to 1369 primary care physicians in active practice in Washington State in 1994. Response rate to the survey was 63%. Chi-square tests and multivariate logistic regression analysis were used to examine the effects of physician characteristics on physicians' self report of use of the PSA test for screening asymptomatic male patients, aged 50 to 80, for prostate cancer. RESULTS: Of the 714 physicians included in the analysis, 68% reported routine use of PSA. Use of PSA varied among physicians on the basis of practice setting, years since medical school graduation, and whether compensation was fee-for-service or salaried. Male physicians trained before 1974 and physicians receiving fee-for-service were significantly more likely than other physicians to recommend screening by PSA. CONCLUSIONS: Results suggest that physicians' personal characteristics such as year of medical school graduation, gender, and mode of reimbursement are related to self-reported PSA use.

An empirical study of the effect of the control rate as a predictor of treatment efficacy in meta-analysis of clinical trials.

If the control rate (CR) in a clinical trial represents the incidence or the baseline severity of illness in the study population, the size of treatment effects may tend to very with the size of control rates. To investigate this hypothesis, we examined 115 meta-analyses covering a wide range of medical applications for evidence of a linear relationship between the CR and three treatment effect (TE) measures: the risk difference (RD); the log relative risk (RR), and the log odds ratio (OR). We used a hierarchical model that estimates the true regression while accounting for the random error in the measurement of and the functional dependence between the observed TE and the CR. Using a two standard error rule of significance, we found the control rate was about two times more likely to be significantly related to the RD (31 per cent) than to the RR (13 per cent) or the OR (14 per cent). Correlations between TE and CR were more likely when the meta-analysis included 10 or more trials and if patient follow-up was less than six months and homogeneous. Use of weighted linear regression (WLR) of the observed TE on the observed CR instead of the hierarchical model underestimated standard errors and overestimated the number of significant results by a factor of two. The significant correlation between the CR and the TE suggests that, rather than merely pooling the TE into a single summary estimate, investigators should search for the causes of heterogeneity related to patient characteristics and treatment protocols to determine when treatment is most beneficial and that they should plan to study this heterogeneity in clinical trials.

Combining single patient (N-of-1) trials to estimate population treatment effects and to evaluate individual patient responses to treatment.

When treating individual patients, physicians may face difficulties using the evidence from center-based randomized control trials (RCTs) due to limitations in these studies generalizability. Therefore, they often perform their own "informal" tests of treatment effectiveness. Single patient ("N-of-1") trials provide a structured design for more rigorous assessment of medical treatments of chronic diseases, but are applied only to the index patient. We present a hierarchical Bayesian random effects model to combine N-of-1 studies to obtain an estimate of treatment effectiveness for the population and to use this population information to aid in the evaluation of an individual patient's trial results. The model's treatment effect estimates are adjustments between the population estimate and the individual's observed results. This adjustment is based upon the within-patient and between-patient heterogeneity. We demonstrate this patient-focused method using published data from 23 N-of-1 trial results comparing amitriptyline and placebo for the treatment of fibromyalgia.

The population risk as an explanatory variable in research synthesis of clinical trials.

The population risk, for example the control group mortality rate, is an aggregate measurement of many important attributes of a clinical trial, such as the general health of the patients treated and the experience of the staff performing the trial. Plotting measurements of the population risk against the treatment effect estimates for a group of clinical trials may reveal an apparent association, suggesting that differences in the population risk might explain heterogeneity in the results of clinical trials. In this paper we consider using estimates of population risk to explain treatment effect heterogeneity, and show that using these estimates as fixed covariates will result in bias. This bias depends on the treatment effect and population risk definitions chosen, and the magnitude of measurement errors. To account for the effect of measurement error, we represent clinical trials in a bivariate two-level hierarchical model, and show how to estimate the parameters of the model by both maximum likelihood and Bayes procedures. We use two examples to demonstrate the method.