Screening geographic areas for unusual survival experience or stage at diagnosis with application to breast and colon cancer

An approach was devised for the problem of detecting geographic areas with poor survival for cancer patients. Methods were proposed to deal with the two major sources of variability in summary survival measures for geographic areas: the distribution of prognostic factors for individuals within an area and random variation. In addition, small geographic areas often had to be combined so that a sufficient number of cases could be obtained in each area for calculation of survival statistics, and a procedure for doing so was developed. Then stage-specific and overall 5-year survival estimates for each geographic area were converted to standardized normal deviates so that outlying observations could be detected using the theory of normal-order statistics. A similar approach was used to determine which geographic areas had unusually high proportions of advanced disease at diagnosis. Our outlier detection procedure was designed for screening available data for geographic areas possibly deserving further study rather than for concluding that survival or staging was substandard in those areas. Methods were applied to data from the Surveillance, Epidemiology, and End Results (SEER) Program for breast and colon cancer. For each cancer site, some geographic areas were identified with unusual survival or stage distributions.

Applicability of sample size calculations based on a comparison of proportions for use with the logrank test

The asymptotic relative efficiency of a test of proportions versus the logrank test is calculated for various clinical trial designs that are used to compare the survival of two treatment groups. The asymptotic relative efficiency is shown to be a reasonable guide to the relative sample sizes required for the logrank and proportions test. It is shown that the efficiency of the proportions test is near 1.0 for designs corresponding to typical studies of cardiovascular disease, for which the duration of the experiment is short compared to mean survival. This result is of practical importance, because sample size calculations based on the comparison of proportions are available to cover many contingencies, including a delay in the onset of full treatment effectiveness, whereas similar calculations have not been published for the logrank statistic. On the other hand, the efficiency of the proportions test can drop to 72% or less for trials in which the accrual period exceeds the mean survival, as is often the case in trials to treat cancer. In such cases, sample size calculations for the proportions test would be [(1/0.72) - 1] = 39% larger than required for the logrank test. Thus, power calculations specifically tailored to the logrank test should be used for studies with a duration comparable to mean survival, if one intends to employ the logrank statistic.

Prognostic factors in patients with resected stage I non-small cell lung cancer. A report from the Lung Cancer Study Group

The authors present prognostic information on recurrence and survival for resected Stage I lung cancer patients with squamous cell carcinoma, adenocarcinoma or large cell carcinoma. The data derive from 392 carefully staged patients and include results from the history and physical examination, preoperative laboratory tests, nature of the surgery, complications, initial pathologic findings following surgical resection, and final pathologic review. A simple multivariate model of recurrence, which is used to classify patients into low, intermediate, and high-risk groups, is based on tumor size and location (T1, T2), histologic type (squamous, nonsquamous/mixed) and nodal status (N0, N1). To model survival, the performance status and the presence of empyema, pneumonia, or wound infection were added to the previous factors. Not all factors associated with increased mortality are associated with increased risk of recurrence, and, in particular, postoperative empyema, pneumonia or wound infections carry an increased risk of death only. Serial measurements of performance status and leukocyte count have the potential for monitoring for increased risk of recurrence and death.

Biased selection of controls for case-control analyses of cohort studies

It is known that unbiased estimates of the relative risk in a cohort study may be obtained by a matched case-control analysis that compares each case with a random sample of controls obtained from those at risk at the time of case incidence. Through inadvertence , or for practical or scientific reasons, a biased referent group may be selected instead. Three kinds of biasing restrictions on controls are commonly imposed: (i) the requirement that controls remain completely disease-free for a fixed time interval, (ii) the exclusion of all cases incident during observation as controls, and (iii) the exclusion, from the referent group, of subjects who develop other diseases, which may be related to the exposure of interest. The bias in estimation of the relative risk associated with each of these restrictions is evaluated under the proportional-hazards model. For several examples of cancer mortality data, the bias from (iii) appears quite small, whereas the bias from (i) can be appreciable and is mostly attributable to the bias from case exclusion (ii). The effect of random variation in the time of onset of exposure is to reduce these biases.

Collaborative study for the evaluation of multiple simultaneous markers in lung cancer

The search for tumor markers suitable for use in diagnosis and management of patients with lung cancer has met with limited success because each available marker lacks sufficient sensitivity and specificity. We have begun a collaborative study to evaluate multiple simultaneous serum markers to determine whether there is a combination that will increase both sensitivity and specificity to the degree that noninvasive serologic tests might provide a means for managing patients with lung cancer. The National Cancer Institute, Bethesda, Maryland, has established a Serum Bank which can supply identical aliquots of the same serum sample to several different laboratories to perform quantitative assays for a number of markers. The results of these assays will serve as a data base to be analyzed for selection of the optimal grouping of markers for the different types of lung cancer and for evaluation of different biostatistical techniques to maximize the benefit derived by multiple marker determination. The evaluation of multiple markers will be carried out in several separate steps: (a) determination of a combination of markers that discriminate advanced lung cancer from benign lung disease; (b) demonstration of the ability of that combination to define early or localized lung cancer; (c) selection of a combination that best separates lung cancer from cancer of other sites; (d) testing of combined markers for the ability to predict response to therapy, including detection of early recurrence before clinical signs appear; and (e) evaluation of the combination(s) for effectiveness in detecting lung cancer in asymptomatic individuals. The experimental design and selection of markers will be presented and discussed.

Monitoring rules for stopping accrual in comparative survival studies

It is useful to separate the decision to stop accrual from the decision to perform a final data analysis in comparative survival trials. Rules for monitoring survival studies are presented that allow for early termination of accrual if evidence mounts against one therapy and if the decision to stop accrual will not prolong the time to analysis unduly. It is shown that these rules for stopping accrual do not detectably perturb the final analysis, which includes a logrank test of no treatment effect and estimation of the log relative hazard with confidence intervals. These methods take advantage of the reservoir of information represented by patients still at risk at the time accrual ceases. Especially with moderate or rapid accrual rates the average number of patients accrued is often reduced substantially and the time to analysis prolonged only slightly.

Use of circulating pregnancy-specific beta 1 glycoprotein as a marker in carcinoma of the breast in women

Pregnancy-specific beta 1 glycoprotein (SP1) was determined by radioimmunoassay in sera from 27 normal women, 33 women with benign breast disease, and 191 women with carcinoma of the breast, staged for extent of the disease. All diagnostic groups exhibited substantial overlap in SP1 values. Those with benign breast diseases tended to have values at least as high as those with cancer. Normal patients tended to have slightly lower values, but this difference may well have been due to the younger ages of the normal patients in our sample, because SP1 values tended to increase with age. Immunochemical dilutions of SP1 in the serum with the highest value (10.2 ng/ml) did not differ significantly from standard placental SP1.

Age and sex as factors influencing remission duration in African Burkitt's lymphoma

The frequency of Burkitt's lymphoma peaks in children six to nine years of age and is twice as high in males as in females. To determine if age or sex had any relationship to outcome of therapy, the authors examined the experience of 387 patients treated for Burkitt's lymphoma at the Burkitt Tumor Project, Accra, Ghana, between 1966 and 1978. No age or sex relationship could be demonstrated with overall survival. However, among children without central nervous system disease who achieved an initial remission, the relapse rates were higher for those diagnosed initially in the age interval 6-9 years than for those diagnosed at younger or older ages. This pattern was seen for each stage of the disease as well as overall. Sex did not influence relapse rate. The relationship of these findings to the Epstein-Barr virus and malaria, both suspected factors in the etiology of this disease, is discussed, but neither appears to explain our results.

Immunological studies in a double blind randomized trial comparing intrapleural BCG against placebo in patients with resected stage I non-small cell lung cancer

Immunological data were obtained during the course of a randomized trial comparing intrapleural BCG plus oral isoniazid (INH) with intrapleural saline plus oral placebo after resection of stage I non-small cell lung cancer. Immunological testing with a variety of assays was carried out with good standardization among six collaborating laboratories and with good reproducibility within each laboratory. Those patients with larger tumors tended to have higher initial white cell counts. The percentage of lymphocytes in the differential was greatest in those with non-squamous cancer histology. Otherwise, no associations were found between initial immunologic parameters and baseline variables. The main effect of BCG/INH therapy was to cause statistically significant increases in purified protein derivative (PPD) skin test induration and PPD in vitro blastogenesis compared with controls. Other skin tests and in vitro assays increased more in the saline/placebo control group, but these treatment differences were usually not statistically significant. Initial white count and neutrophil count elevations were found to be associated with increased risk of recurrence. Even after adjustment for treatment and tumor stage, initial neutrophil count elevation was associated with increased risk of recurrence. Surprisingly, a low 29 degrees C T cell rosette index was associated with a decreased risk of recurrence, though the differences were minimal. Serial immunological tests were carried out to evaluate their potential for monitoring disease recurrence. White count elevations continued to be significantly associated with increased risk of recurrence, but more follow-up data are needed before other associations can be assessed.

Evaluating serial cancer marker studies in patients at risk of recurrent disease

The proportional hazards model of Cox (1972, Journal of the Royal Statistical Society, Series B 34, 187--220), with a time-dependent covariate, is used to analyze serial cancer marker data. A particular advantage of this method is the case with which missing marker data are handled. Analysis of a real data set shows that high levels of the cancer marker carcinoembryonic antigen (CEA) are associated with increased risk of death in patients with resected colorectal cancer. Several aspects of CEA marker history are analyzed, including CEA level at death time t, CEA level 200 days prior to time t, and whether or not CEA exceeded 5 ng/ml prior to t. Methods to test the hypothesis of no marker effect and to give estimates and confidence intervals for model parameters are outlined both for continuous and for grouped time-to-response data. For grouped data a likelihood ratio test of the proportional hazards assumption is suggested.

Correlative genetic variation in natural populations of cats, mice and men

The study of the extent and basis of gene-enzyme variation has long been a principal concern of population genetics. Numerous surveys have indicated considerable amounts of genetic variation detectable in natural populations, with few exceptions. The variances of average heterozygosities (H) between species and among populations within species are large, prompting Lewontin to emphasize the importance of large gene sample sizes and Selander to encourage analysis of variation of homologous gene-enzyme systems when making species comparisons. We present here a comparative genetic analysis of electrophoretic variation at 57 homologous biochemical loci of cats, mice and men. The distribution of polymorphism among the sampled loci in the three species was nonrandom. A large group of sampled loci (60%) were monomorphic in all three species, whereas a second group (30%) of the loci were polymorphic in two or more species. This conservation of the tolerance of genetic polymorphism is apparently more a characteristic of a particular locus than of the vertebrate species or of the genome. The current hypotheses for classifying polymorphic and monomorphic loci in terms of physiological and physical enzyme characteristics have been re-examined.

An analysis of comparative carcinogenesis experiments based on multiple times to tumor

Methods are given for the comparison of two treatment groups in experiments giving rise to multiple tumors. Methods based on the gaps in time between successive tumors are emphasized, but, for comparison, one method based directly on times to tumor is also presented. When applied to the results of an experimental animal carcinogenesis study, these analyses show that a diet supplemented with retinyl acetate reduces the hazard of mammary tumors, compared with controls, in every gap, and they allow one to combine evidence from the several gaps to obtain a powerful test for treatment effect.

A molecular approach to the identification and individualization of human and animal cells in culture: isozyme and allozyme genetic signatures

The electrophoretic resolution of a group of genetically monomorphic gene-enzyme systems that are developmentally and biologically ubiquitous has been used to provide a species-specific and type-specific biochemical characterization of various cultured cells. The relative mobilities of gene-enzyme systems representing nine distinct gene products from cell cultures of 25 species from Drosophila to man are presented. These isoenzymes effectively discriminate interspecies cell-to-cell contamination and almost invariably serve to identify the contaminating species. The resolution of eight polymorphic gene-enzyme systems in human cell cultures provides a virtually unique allozyme genetic signature as a monitor of intraspecies cellular contamination. The genetic signatures of 47 commonly used human cells are presented. Included in the test were seven putative HeLa (human cervical carcinoma) contaminants each of which expressed a signature identical with that of HeLa. The probability that an unrelated human cell line will have a signature identical to a typed cell is computed for each line from the genotypic frequencies at each locus in a population of cultured human cells. The gene frequencies of this cell population are comparable to the same frequencies in natural human populations. The most common human signature has a frequency (and therefore a probability) of 0.02. The majority of the 17,010 possible signatures are far less probable. A calculation of the theoretical incidence of chance matching of signatures within test groups of two or more individuals is presented. The probability of a chance match between any two randomly selected individuals is 0.004 and among five randomly selected individuals is 0.034. The allozyme genetic signature represents a definitive monitor of cell identity and is presented as a standard of cell and tissue identification for a variety of biological studies.