On the use of historical control data in pre-clinical safety studies.

A number of methods to formally incorporate historical control information in pre-clinical safety evaluation studies have been proposed in literature. However, it remains unclear when one should use historical data. Focusing on the logistic-normal model, we investigate situations where historical studies may prove to be useful. Aspects of estimation (precision and bias) and testing (power) for treatment effect are investigated under different conditions such as the number of historical control studies, the degree of homogeneity amongst them, the level of treatment effect and different control rates. The possibility to use a selected subset of historical control studies is also explored.

Geographical differences in cancer incidence in the Belgian province of Limburg.

Correctly addressing the questions of worried citizens with respect to possible clusters of cancer occurrence requires a risk communication strategy that is informed by a previously established analytical procedure. The aim of this study was to analyse cancer registration data in order to identify municipalities or clusters of municipalities with an increased incidence of one or more cancer types, adjusted for background characteristics at the same level. Ideally, the approach is proactive, straightforward, and easy for untrained citizens to follow and imprecision effects are taken into account. For all municipalities and most cancers, all relevant calculations were performed proactively and all methods and decision thresholds were defined beforehand. For each municipality, standardised incidence ratios (SIRs) were calculated and smoothed using a Poisson-gamma (PG) and a conditional autoregressive (CAR) model. Clusters were confirmed using the Spatial scan statistic of Kulldorff. Identified clusters were tested for possible confounders using all information that was available for each municipality. The Limburg Cancer Registry, serving the population of the Belgian province of Limburg (n=781 759) was used. We identified a possible cluster of increased prostate cancer incidence (smoothed SIRs around 1.2) and a cluster of increased bladder cancer incidence in males that included seven municipalities with CAR-smoothed SIRs between 1.5 and 2.1. SIRs followed a more or less circular decrease around the centre that was situated in Alken and Hasselt, the provincial capital. Bladder cancer incidence was positively related to an index of socio-economic status (SES) per municipality. No relationship was found with the other indexes that were available. 82% of all bladder cancers were transitional cell carcinomas (TCC). A repeated analysis based on TCCs only resulted in similar results with CAR-smoothed relative risks that tended to be even higher in the cluster zone. A pre-emptive analysis of possible cancer incidence clustering on the municipality level proved to be feasible.

Prostate-specific antigen screening coverage and prostate cancer incidence rates in the Belgian province of Limburg in 1996-1998.

According to the 1996-1998 cancer incidence report of the cancer registry of the Belgian province of Limburg (LIKAR), prostate cancer is the most common cancer in men with a crude invasive cancer incidence rate of 123.7 per 100000 person-years (125.4 and 81.8 after standardization for the European and the world standard population). In a study on geographical differences between the occurrence of cancers in municipalities, prostate cancer standardized incidence rates (SIRs) were significantly higher in a number of municipalities, with mean relative risks of 1.2 and 1.3 after full Bayesian smoothing. We hypothesized that prostate cancer incidence rates are largely influenced by the prostate-specific antigen (PSA) screening policy of local physicians and that differences between municipalities are more informative about local screening habits then about real differences in cancer occurrence. The aim of this study was to test this hypothesis by relating local prostate cancer SIRs to the PSA screening coverage of the population of men in each municipality. The SIRs of prostate cancer in 1996-1998 for each municipality were provided by LIKAR. They related to all histologically or cytologically proven new invasive prostate cancers during these years. For each municipality, PSA coverage data were provided by the largest sick fund of the region. Coverage was defined as the proportion of men above the age of 40 that was tested at least once within the registration period. The SIR of each municipality (dependent variable) was related to the age-standardized corresponding coverage (independent variable) by linear regression and was adjusted for the number of inhabitants per municipality: log (standardized incidence rate) = 164 + 602 * (standardized PSA coverage), = 0.12. The model explained 6% of the variance in incidence. In conclusion, in this study no statistically significant relationship was identified between PSA coverage and prostate cancer incidence rate per municipality. This could result from no such relationship existing or from low statistical power.

Pesticides and mortality from hormone-dependent cancers.

Endocrine-disrupting chemicals are considered to be a possible cause of hormone-dependent cancers. In areas of high exposure to pesticides, people are concerned about the long lasting toxicity of pesticides, some of which are possibly hormonally active. We collected for each Belgian municipality (n = 589) the latest mortality statistics from breast and prostate cancer (period 1985-1994) and the latest data on crops and pesticides (1998). In addition, data on possible confounders such as population density, degree of urbanization, industrial activity and the presence of an incinerator were collected as well. The data were analysed with spatial statistics that takes into account the spatial nature of the data. There is a large variation in crops and pesticide exposure among the municipalities, the highest exposure being seen in the fruit production area. Apart from use of defoliants and potato cultivation, no consistent correlation was detected between crops, pesticides and mortality from breast and prostate cancer. Our data cannot support the hypothesis of a relationship between total and class-related pesticide use and breast and prostate cancer mortality. However, the increased mortality due to breast cancer and to a lesser extent due to prostate cancer in traditional potato-growing areas needs attention and more research.

Evaluation of surrogate endpoints in randomized experiments with mixed discrete and continuous outcomes.

A statistical definition of surrogate endpoints as well as validation criteria was first presented by Prentice. Freedman et al. supplemented these criteria with the so-called proportion explained. Buyse and Molenberghs pointed to inadequacies of these criteria and suggested a new definition of surrogacy based on (i) the relative effect linking the overall effect of treatment on both endpoints and (ii) an individual-level measure of agreement between both endpoints. Using data from a randomized trial, they showed how a potential surrogate endpoint can be studied using a joint model for the surrogate and the true endpoint. Whereas Buyse and Molenberghs restricted themselves to the fairly simple cases of jointly normal and jointly binary outcomes, we treat the situation where the surrogate is binary and the true endpoint is continuous, or vice versa. In addition, we consider the case of ordinal endpoints. Further, Buyse et al. extended the approach of Buyse and Molenberghs to a meta-analytic context. We will adopt a similar approach for responses of a mixed data type.

The validation of surrogate endpoints in meta-analyses of randomized experiments.

The validation of surrogate endpoints has been studied by Prentice (1989). He presented a definition as well as a set of criteria, which are equivalent only if the surrogate and true endpoints are binary. Freedman et al. (1992) supplemented these criteria with the so-called 'proportion explained'. Buyse and Molenberghs (1998) proposed replacing the proportion explained by two quantities: (1) the relative effect linking the effect of treatment on both endpoints and (2) an individual-level measure of agreement between both endpoints. The latter quantity carries over when data are available on several randomized trials, while the former can be extended to be a trial-level measure of agreement between the effects of treatment of both endpoints. This approach suggests a new method for the validation of surrogate endpoints, and naturally leads to the prediction of the effect of treatment upon the true endpoint, given its observed effect upon the surrogate endpoint. These ideas are illustrated using data from two sets of multicenter trials: one comparing chemotherapy regimens for patients with advanced ovarian cancer, the other comparing interferon-alpha with placebo for patients with age-related macular degeneration.