Methodology for lognormal modelling of malignant pleural mesothelioma survival time distributions: a study of 5580 case histories from Europe and USA.

A truncated left-censored and right-censored lognormal model has been validated for representing pleural mesothelioma survival times in the range 5-200 weeks for data subsets grouped by age for males, 40-49, 50-59, 60-69, 70-79 and 80+ years and for all ages combined for females. The cases available for study were from Europe and USA and totalled 5580. This is larger than any other pleural mesothelioma cohort accrued for study. The methodology describes the computation of reference baseline probabilities, 5-200 weeks, which can be used in clinical trials to assess results of future promising treatment methods. This study is an extension of previous lognormal modelling by Mould et al (2002 Phys. Med. Biol. 47 3893-924) to predict long-term cancer survival from short-term data where the proportion cured is denoted by C and the uncured proportion, which can be represented by a lognormal, by (1 - C). Pleural mesothelioma is a special case when C = 0.

Methodology to predict a maximum follow-up period for breast cancer patients without significantly reducing the chance of detecting a local recurrence.

For breast cancer where the prognosis of early stage disease is very good and even when local recurrences do occur they can present several years after treatment, the hospital resources required for annual follow-up examinations of what can be several hundreds of patients are financially significant. If, therefore, there is some method to estimate a maximum length of follow-up Tmax necessary, then cost savings of physicians' time as well as outpatient workload reductions can be achieved. In modern oncology where expenses continue to increase exponentially due to staff salaries and the expense of chemotherapy drugs and of new treatment and imaging technology, the economic situation can no longer be ignored. The methodology of parametric modelling, based on the lognormal distribution is described, showing that useful estimates for Tmax can be made, by making a trade-off between Tmax and the fraction of patients who will experience a delay in detection of their local recurrence. This trade-off depends on the chosen tail of the lognormal. The methodology is described for stage T1 and T2 breast cancer and it is found that Tmax = 4 years which is a significant reduction on the usual maximum of 10 years of follow-up which is employed by many hospitals for breast cancer patients. The methodology is equally applicable for cancers at other sites where the prognosis is good and some local recurrences may not occur until several years post-treatment.