Search

Background: USON 01062 (O’Shaughnessy J, et al. Proc SABCS, 2010, abst S4-2) showed no improvement in the primary endpoint of disease-free survival (DFS) (median FU 5 yrs: HR 0.84, 95% CI: 0.67-1.05; p = 0.125) with the addition of capecitabine (X) to standard adjuvant chemotherapy, but showed improvement in OS (HR 0.68, 95% CI: 0.51-0.92; p = 0.011). Exploratory analysis of local pathology-assessed Ki67 suggested benefit from adjuvant X in pts with more highly proliferative cancers with Ki67 ≥ 10% (Pippen J et al. Proc ASCO, 2011, abst 500). The objective of this study is to determine whether centrally-performed Ki67 IHC results corroborate or refute this finding. Methods: 2610 pts with resected high risk EBC were randomized to receive 4 cycles of AC (doxorubicin 60mg/m2 and cyclophosphamide 600mg/m2) IV every 3 wks for 4 cycles followed by either docetaxel 100mg/m2 IV or docetaxel 75mg/m2 IV plus X 825mg/m2 PO bid for 14 days every 3 wks for 4 cycles. Archival primary breast cancer tissue was collected on 2000 pts for predictive biomarker analyses. Central Ki67 IHC was performed using the anti-Ki67 monoclonal antibody SP6 and was read by one pathologist (HK) according to published recommendations (Dowsett M, et al. JNCI 103:1-9, 2011). Results: Central Ki67 IHC has been performed on 1440 pts who had centrally-validated informed consents. The distribution of% Ki67-positive cells by locally-assessed ER/HER2 subtype is shown below. 45% of HR+ HER2- BCs had a Ki67 ≤ 10%, while 24% had a Ki67 11% to 20%, and 31% had a Ki67 > 20%. The concordance between the local vs central Ki67 results was low at 46% for Ki67 20%. The central Ki67 results tended to be higher than the local testing results. Central mRNA classifiers were developed for ER, PR, HER2 and Ki67 using Fluidigm Microfluidics Dynamic Arrays and correlate highly with central IHC assessment of these markers. Conclusions: HR+ HER2- EBC is enriched for cancers with a low proliferative rate, a group of pts unlikely to benefit from the cell cycle-specific cytotoxic agent, capecitabine. Analyses of the impact of adjuvant X added to AC/T in EBC pts according to ER status, and according to Ki67 (analyzed as a binary and continuous variable) will be performed prior to SABCS, 2013. Number of Patients% Ki67 Pos CellsTotal *HR+TNHER2+/HR+HER2+/ HR-0-104163622222711-151391066151016-20126871615821-3018411539201031-1005751403423555Total144081042510790*Totals do not equal sum of subtype categories due to missing HER2 information Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P6-09-01.

Populations of the green frog (Rana clamitans) and the mink frog (Rana septentrionalis) in Algonquin Provincial Park, Ontario, Canada, were studied from May 1985 through October 1987 and from May 1991 through September 1993. Parameters estimated from fitting logistic curves to growth data obtained from marked individuals were used to test for interspecies and intersexual differences in growth trajectories. Female R. septentrionalis had a significantly greater asymptotic length than males. We found no sexual size dimorphism in R. clamitans. Asymptotic size was significantly greater for R. clamitans than for R. septentrionalis. Age was estimated from a combination of extrapolation from size-frequency data and from production of size-specific growth curves from recapture data. Age-specific size and growth rate were measured to test the hypothesis that differences in growth trajectories could be predicted from differences in patterns of survivorship. Both R. septentrionalis and R. clamitans were larger at transformation, and older at maturity, than has been reported in previous studies. Although significantly larger at transformation, R. septentrionalis was significantly smaller than R. clamitans in all subsequent age-classes. The growth rate of R. clamitans was significantly greater than that of R. septentrionalis for all age-classes except 0 years post transformation. The observed differences in growth trajectories and age at maturity were consistent with predictions based on differences in the pattern of survivorship.

Contemporary geographical range and patterns of genetic diversity within species reflect complex interactions between multiple factors acting across spatial and temporal scales, and it is notoriously difficult to disentangle causation. Here, we quantify patterns of genetic diversity and genetic population structure using mitochondrial DNA sequences (101 individuals, cytochrome b) and microsatellites (816 individuals, 12 loci) and use Approximate Bayesian computation methods to test competing models of the demographic history of eastern and western foxsnakes. Our analyses indicate that post-glacial colonization and past population declines, probably caused by the infilling of deciduous forest and cooler temperatures since the mid-Holocene, largely underpin large-scale genetic patterns for foxsnakes. At finer geographical scales, our results point to more recent anthropogenic habitat loss as having accentuated genetic population structure by causing further declines and fragmentation.

[Read before The Royal Statistical Society at a meeting organized by the Research Section on Wednesday, October 25th, 1989, Professor D. V. Hinkley in the Chair] SUMMARY The paper addresses the evergreen problem of construction of regressors for use in least squares multiple regression. In the context of a general sequential procedure for doing this, it is shown that, with a particular objective criterion for the construction, the procedures of ordinary least squares and principal components regression occupy the opposite ends of a continuous spectrum, with partial least squares lying in between. There are two adjustable 'parameters' controlling the procedure: 'alpha', in the continuum [0, 11, and 'omega', the number of regressors finally accepted. These control parameters are chosen by crossvalidation. The method is illustrated by a range of examples of its application.

The spectral transmission and electron emissivity responses, measured for a series of typical photocathodes, are presented and analysed. Specifically, samples of S1, S20, S25, Bialkali and two types of solar-blind telluride photocathodes were investigated in both transmission and reflection modes of operation. The transmission mode is more convenient for imaging, night vision and for scintillation counting applications such as CT scanners and is more commonly used than the reflection mode. However, more recent work has focussed on the reflection photocathode as a source of electrons with low energy spread used for electron guns for microscopy and lithographic free electron lasers [1]. Our analysis provides a determination of the reflectivity of the substrate/cathode and cathode/vacuum interface, enabling the refractive index to be deduced. The high apparent quantum efficiency (QE) of some conventional photocathodes is shown to be due to the conversion of each photon to two or more electrons.

Common snapping turtle (Chelydra serpentina serpentina) eggs from five sites within the Great Lakes basin, and from a reference site in north-central Ontario were collected during 1981-1991 and analyzed for four organochlorine pesticides, polychlorinated biphenyls (PCBs) including six non-ortho PCBs, polychlorinated dibenzodioxins (PCDDs), and polychlorinated dibenzofurans (PCDFs). The pattern of geographic variation was consistent over time in eggs with Cootes Paradise/ Hamilton Harbour and Lynde Creek eggs on Lake Ontario containing the highest concentrations and most PCDD and PCDF congeners among all sites. Eggs from Cranberry Marsh on Lake Ontario contained organochlorine concentrations similar to those from Big Creek Marsh and Rondeau Provincial Park on Lake Erie except PCDDs and PCDFs which occurred at higher concentrations and more congeners were detectable in Cranberry Marsh eggs. Concentrations of most contaminants in turtle eggs from Algonquin Park, the reference site, have significantly decreased in the past decade. Dieldrin concentrations, however, increased in Algonquin Park eggs from 1981 to 1989. Significant decreases in concentrations of hexachlorobenzene, mirex and PCBs occurred between turtle eggs collected in 1981/84 and 1989 at Big Creek Marsh and Rondeau Provincial Park, whereas there was no significant change in concentrations of p,p'-DDE and dieldrin. In Lake Ontario eggs, concentrations of PCBs, p,p'-DDE and dieldrin increased significantly between 1984 and 1991. Differences were also found in patterns of temporal variation in contamination between herring gulls (Larus argentatus) and snapping turtles which were attributed to differences in diet. Elevated and continued contamination in turtle eggs from Lake. Ontario is probably due to a combination of local sources of chemicals and consumption of large migratory fish that spawn in wetlands inhabited by these turtles.

To determine the longitudinal cost of the treatment of patients with breast cancer.An analysis was performed of 200 women with 205 newly diagnosed breast cancers during 1989 in a health maintenance organization population (US Healthcare, Blue Bell, Pa). Medical records and claims data were analyzed for the total costs of medical care during the 4-year period after diagnosis. The costs over time were analyzed for clinical stage and use of mammography screening.The total costs of medical care during the 4-year period after diagnosis were strongly related to clinical stage at diagnosis, with higher total costs for patients with stages III to IV at diagnosis compared with patients with stages 0 to II at diagnosis. The cost for all stages of disease declined after years 1 to 2, with the exception of stage II, which increased slightly in years 3 to 4. The use of screening mammography was associated with a significant decrease in the cost of medical care during the 4-year study period.The goal of mammography screening programs should be to achieve downstaging to stages 0 to 1 to achieve reduction in breast cancer mortality and to reduce the overall consumption of health care resources for the treatment of breast cancer. These cost data should be considered within the framework of future cost-effective analysis for screening mammography programs.

This is a preliminary study of die reactions of mercury (Hg) in the human mouth with dental amalgams. It was conducted by analysing saliva samples from subjects with amalgam fillings and control subjects with no amalgams. Samples were collected both prior to and after cleaning the mouth. These samples were analyzed for elemental mercury (Hg°), inorganic mercury (Hg2+) and methylmercuiy (MeHg). We concluded that the concentrations after cleaning represented the systemic concentrations. Hg2+ and MeHg were found in all systemic samples from both subjects and controls, while Hg° was found only in the samples from subjects with amalgams. In the control group, the concentrations found before and after cleaning die mouth were equivalent In the amalgam group, concentrations of Hg2+ found before cleaning the mouth were 10 to 40 times higher than those found after cleaning, suggesting that die oxidation reaction of Hg° into Hg2+ takes place. For MeHg, a similar but less pronounced pattern as Hg2+ was found, supporting methylation in the mouth.

547 Background: Randomized phase III USON 01062 trial determining if patients with EBC would benefit from addition of capecitabine to docetaxel after AC (AC-T vs AC-XT). AC-T: docetaxel 100mg/M2 IV; AC-XT: docetaxel 75mg/M2 IV with capecitabine 825mg/M2 PO BID 14/7 days every 21days for 4 cycles. The primary endpoint, improvement in disease-free survival (DFS) at 5 years, was not met (HR=0.84, p=0.12) likely due to lower-than-expected event rate. The secondary endpoint, overall survival (OS), was improved with capecitabine (HR 0.68, p=0.01) (O’Shaughnessy, J. ASCO, 2011). Methods: Molecular analyses demonstrate that pleomorphic lobular (mixed lobular/ductal) carcinomas evolve from the same precursor and/or through the same genetic pathway as classical lobular cancers (Reis-Filho, J., J Path, 2005). We conducted exploratory analyses to evaluate the addition of adjuvant capecitabine in ductal vs lobular or lobular/ductal (mixed) EBC within USON 01062. Histology was classified according to local pathology assessment on patients’ primary cancers. Results: In ductal patients (n=2195), there was no difference in DFS (HR=0.92, p=0.48) and OS (HR=0.75, p=0.07) with AC-T vs AC-XT. In lobular/mixed patients (n=355), adding capecitabine improved DFS (HR=0.55, p=0.055) and OS (HR=0.38, p=0.04). There was no difference in DFS (HR=1.004, p=0.98) in the ER+ ductal patients (n=1258) with the addition of capecitabine. Conclusions: Ductal and lobular cancers have distinct histologic and molecular characteristics; lobular cancers are generally less sensitive to chemotherapy (Cristofanilli, M. JCO, 2005). This exploratory analysis suggests that patients with lobular/mixed EBC may benefit from adjuvant capecitabine. This hypothesis requires evaluation in other adjuvant capecitabine trials. [Table: see text]

Statistical analyses were used to determine relationships between body size, clutch size and mass, and relative clutch mass and levels of organochlorine pesticides and seven polychlorinated biphenyl congeners in the eggs of adult female common snapping turtles (Chelydra serpentina serpentina). No significant correlations were found between body size (body mass, carapace length, carapace width, plastron length) and lipid normalized concentrations of p,p′-DDE, mirex, dieldrin, and polychlorinated biphenyl (PCB) congeners (IUPAC): #52, #105, #118, #138, #153, #180, #194, and the sum concentration of those congeners. Small sample size and clumping of data around the mode of the body size values prevented inferences of nonlinear relationships. It was concluded that body size and clutch characteristics are not strong or reliable predictors of the level of contaminants in snapping turtle eggs and that adjustment for those parameters would not reduce variation in contaminant levels among clutches. Other variables such as individual food preferences and/or foraging activities are more likely to cause variation in chemical concentrations among clutches of eggs within a population. In order to reduce interclutch variation in contaminant levels to 38.6–55.9% in snapping turtles, sample sizes of at least 15 clutches per site are recommended.

500 Background: First results of study USON 01062 comparing adjuvant AC→T vs AC→XT in high-risk EBC [O’Shaughnessy, SABCS 2010] showed the primary endpoint, improvement in DFS, was not met (median ...

Background: Combination regimens incorporating anthracyclines and taxanes are among the most effective for EBC and are particularly suitable for pts with high-risk disease. Significant efficacy benefit has been shown in phase III trials integrating capecitabine (X) into anthracycline/taxane-containing (neo)adjuvant regimens (Joensuu H, et al. Lancet Oncol 2009; Steger G, et al. ASCO 2010). We present first efficacy results of a large, randomized, multicenter phase III study comparing adjuvant doxorubicin plus cyclophosphamide (AC) followed by docetaxel (T) with or without X in high-risk early stage BC. Methods: Pts aged 18-70 years, with high-risk, histologically-confirmed BC were eligible. High risk was defined as ≥1 positive lymph node, T1-3, and M0; or node negative with tumors >2 cm and M0; or node negative with tumors >1 cm, both ER and PgR negative, and M0. Treatment comprised four 3-weekly cycles of AC (A: 60mg/m2, C: 600mg/m2, both day 1) followed by four 3-weekly cycles of T (100mg/m2 day 1) or XT (X: 825mg/m2 b.i.d., days 1-14; T: 75mg/m2 day 1). Pts with hormone receptor-positive disease received tamoxifen or aromatase inhibitor for 5 years, and after 2005, pts with HER2-positive disease were offered 1-yr of concurrent or post-study trastuzumab. Primary endpoint: DFS (518 events were expected at 5 years); secondary endpoints: OS; safety; delivered dose intensity. Results: Between Aug 2002 and Feb 2006, 2,611 pts were randomized to AC→T (n=1,304) or AC→XT (n=1,307). The treatment arms were well balanced at baseline: median age was 51 yrs (range 26–72) and most pts had ECOG PS 0 (91%). The study failed to meet its primary endpoint of DFS (HR 0.84, 95% CI: 0.67-1.05; p=0.125) after a median follow-up of 5 years, with 304 events. However, a statistically significant improvement in OS was seen in pts receiving AC→XT vs AC→T (HR 0.68, 95% CI: 0.51-0.92; p=0.011), with 183 events. Subgroup analyses of DFS and OS appeared to favor the AC→XT arm over the AC→T arm, with few exceptions. The frequency of AEs was similar in both arms: 99.8% AC→T (n=1,305) vs 100% AC→XT (n=1,283), as was the incidence of serious AEs: 20.2% vs 15.6%, respectively. Differences were noted between the AC→T and AC→XT arms, respectively, in terms of the incidence of grade 3 hand-foot syndrome (3.8% vs 18.1%), and grade 3/4 stomatitis (4.5% vs 9.1%), diarrhea (2.9% vs 5.1%) and febrile neutropenia (13.1% vs 9.4%). Median dose intensity of T was 0.97 (range: 0.08-1.41) in the AC→T arm and 0.96 (range: 0.03-1.45) in the AC→XT arm; the corresponding value for X was 0.67 (range: 0.00-1.20), which is lower than reported with XT in MBC. Conclusions: Although this pivotal study failed to meet its primary endpoint of DFS, improvements in OS were seen with the addition of X to a standard anthracycline/taxane-containing adjuvant regimen. These results must be interpreted with caution due to the lower than expected event rate at 5 years. No new safety signals were detected, but the incidence of grade 3/4 stomatitis in the AC→XT arm was higher than previously reported with XT in the metastatic setting. Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr S4-2.

In some samples of pig litters, the number of males per litter exhibits significant sub-binomial dispersion: there are fewer unisexual and more "sex-balanced" litters than expected under a binomial model. It is argued here that the explanation is that the sex of zygotes may be subject to hormonal control with the consequence that, within a litter, the zygotes do not all have the same probability of being male. This variation in probability is approximated by a quadratic function of the time during oestrus that a zygote is formed so as to parallel the variation in maternal hormone levels. Other explanations are examined briefly, but each seems to have serious weaknesses.

516 Background: We assessed 10 menopausal symptoms at baseline and every 3 months during the first year of an ongoing randomized Phase III trial comparing relapse-free survival of postmenopausal women with receptor positive, early breast cancer treated with exemestane (an aromatase inactivator) or tamoxifen for 5 years. Methods: Symptoms were assessed by each patient as none, mild, moderate, or severe with more detail used to establish a “hot flash score”. The median age of 997 evaluable pts was 65 years (range, 40–90). The symptoms were examined statistically by analysis of variance with repeated measurement design. Results: Vaginal dryness (p=0.0021) and bone/muscle aches (p

SIUMMARY A Bayesian decision theory approach to the choice of regression design is considered when it is intended to use the regression to help control the dependent variable at a chosen value, the control to be effected by choosing certain of the independent variables and fixing them at selected values. Some key word8: Bayesian decision theory; Control; D-optimum; Linear regression; Preposterior analysis; Regression design. 1. INTRODIUCTION This paper is concerned with the design of a regression experiment in which observations of a dependent variable are obtained at selected values of a set of independent variables. Using Bayesian decision theory, Lindley (1968) considered the analysis of data from a regression experiment in two situations, namely when it is desired to choose the best subset of independent variables (a) to predict a future value of the dependent variable, (b) to control the level of the dependent variable at a preassigned value. The first situation he termed a 'prediction problem' and the second a 'control problem'. The optimal choice of a designed regression experiment for Lindley's approach to the prediction problem was considered by Brooks (1972). In the present paper, the optimal choice of designed experiment for the control problem is considered, whe-n the regression is linear in the independent variables. To do this, a decision-theoretic analysis of the following sequence of decisions and events is considered. First, the size n of the experiment is decided, and then n sets of values X1, ..., X

SUMMARY The expected gain in information provided by an experiment is used to investigate the loss of information when observations on lifetimes in reliability and survival studies are censored. It is assumed that the lifetime distribution is exponential. The loss of information through only obtaining data at given time points is also considered.

The s-expected gain in information provided by an experiment is used to help assess the loss of s-information through censoring in reliability studies. Large sample, relative s-efficiencies of censored studies compared to complete studies are calculated assuming the lifetime distribution is Weibull. For example, when q, the proportion (if type II censoring) or s-expected proportion (if type I censoring) of complete lifetimes to be censored, is up to about 20%, a censored study requires roughly 200q% extra observations for the same s-expected gain in s- information as a complete study. The results allow an ordering of studies with different sample sizes and stopping points with respect to their s- expected gains in s-information.

This study describes the sequential formation of lesions associated with the endogenous development of Sarcocystis rauschorum (Apicomplexa: Sarcocystidae) in varying lemmings, Dicrostonyx richardsoni. Lethal doses of sporocysts (greater than 500) were orally administered to lemmings examined 1-6 days postinoculation (DPI) whereas sublethal doses were administered to lemmings examined subsequently. Transient necrosis and purulent inflammation, in association with precystic merogony, occurred in the liver by 4.5 DPI, peaked at 6 DPI and subsided beginning at 11 DPI with the liver returning to normal by 15 DPI. Cyst formation in skeletal and cardiac muscle was associated with purulent inflammation and sarcolemmal proliferation beginning at 9 DPI. These lesions persisted to 42 DPI. In addition, multifocal nonsuppurative meningoencephalitis was present in six of 11 infected lemmings examined between 11 and 15 DPI.

The river Laerdalselva, West Norway, was regulated in the autumn of 1974. Regulation led to an increase in winter flow and a decrease in summer flow in a section where there was natural production of salmon. A slight decrease in summer temperature was also recorded in the uppermost part of this section. No data existed on juvenile Atlantic salmon before regulation and the regulation effect on juvenile fish population parameters is therefore based on samples taken from adult salmon in the period 1969 to 1984. No differences in growth, smolt age, and smolt size which could be related to the regulation of the river were found. The smallest mean sizes of yearlings (0+) were found in 1964 and 1967. After regulation the mean size was never lower than before. Mean smolt age was between 3–1 and 3–9 years, but after regulation never exceeded that found before. River growth was slow, three year old smolts growing faster than four year old smolts. However, there were no differences in river growth before and after regulation.

In adult male brown lemmings, Lemmus trimucronatus, a specialized pad of sebaceous glands forms the perineal gland, consisting primarily of two longitudinal ridges lateral and parallel to the midventral line of the scrotum. The ridges extend from the posterior margin of the penis sheath to the anterior edge of the rectum, and correspond externally with a pair of longitudinal furrows. Externally, the gland is characterized by brown pigmentation and coarse, transparent hairs. Histologically it is comparable to, though larger than, the rump gland of the same species. Development of the perineal gland is correlated with attainment of sexual maturity.On the basis of other studies analogous skin glands in other rodents, a scent-producing function is hypothesized for the perineal gland.

Mitomycin C together with either vindesine or vinblastine was given to 48 patients with previously treated advanced breast cancer. Thirteen (35%) of the 37 evaluable patients had a complete (one patient) or partial (12 patients) response. Overall median duration of response was 189 days (range, 90-700 days). Fifteen patients received mitomycin C and vindesine with six responses (40%) and a median response duration of 247 days (range, 162-700 days). Twenty-two patients received mitomycin C and vinblastine with seven responses (32%) and median response duration of 164 days (range, 90-330 days). Response duration for patients treated with mitomycin C plus vindesine was longer than that associated with mitomycin C plus vinblastine (p = 0.09). Significant toxicity included myelosuppression and neurologic symptoms, but was uncommon (less than 10% of patients). Therefore, the combination of mitomycin C and a vinca alkaloid appears to be useful in far-advanced refractory breast cancer.

Using Bayesian decision theory, the problem of how to choose an optimal regression design is considered when there is prior knowledge about the regression parameters available and the data are to be analyzed so that a future value of the dependent variable can be predicted.

SUMMARY The choice of ratio of sample sizes of two groups of binary observations for comparing two proportions is studied using a Bayesian approach. Prior uncertainty about the unknown proportions thus enters the design calculations.

Eleven patients with mature or peripheral T-cell lymphoma (PTL), other than mycosis fungoides, were identified using an extensive battery of T- and B-cell markers. Eight cases had a histologic diagnosis of either diffuse large cell or mixed lymphoma, three of small cell type. All cases had one or more “mature” T-antigens and an absence of B- and immature T-antigens. Assessment of T-antigens included E-rosettes (Er), anti-Leu 1-7 and Tdt. The authors delineated striking heterogeneity of T-antigen expression: 9 different immunotypes in 11 cases. Subset T-antigen assessment indicated T-helper neoplastic cells in five cases and T-suppressor in two. The remaining four had universal T-antigens alone. Seven cases appeared to have “novel” T-phenotypes hot corresponding to normal T-ontogeny phenotypes. Novel or idiosyncratic phenotypes may be a key characteristic of FTL. Since no single T-antigen, including Er and Er receptor (Leu-S), was expressed in all cases, a battery of monoclonal antibodies is. necessary to detect PTL. Clinically, the authors found PTL⋅ unexpectedly aggressive, despite mature immunotype. Most patients were elderly (median age 69); all had extranodal disease with cutaneous involvement (six cases) most frequent. Responses to chemotherapy frequently proved transient, with median survival of nine months. A fulminant course was noted even with localized presentation. Clinical outcome suggests PTL requires new therapeutic strategies.

Summary Two young patients with head injuries subsequently developed neurogenic pulmonary oedema. The origin and pathways of the reflex response to cerebral trauma are discussed, with emphasis on the role of raised intracranial pressure and the sympathetic division of the autonomic nervous system in precipitating movement of systemic circulatory volume into the pulmonary circulation, Therapeutic measures are discussed which correct the progression of this condition by reducing intracranial pressure and blocking the systemic effects of autonomic reflex activity. This leads to the basis of a simple regime to treat future cases.

We will be considering what Lindley (1968) in his paper on the choice of variables in multiple regression termed 'a prediction problem'. By this we mean that data from a regression experiment is analyzed in such a way that we can predict a future value of the dependent variable and choose which independent variables to use for the prediction. In this paper, the regression experiment is designed, i.e. the levels of the independent variables are selected. Our object is to find the optimal design. Although many papers have been published on obtaining optimal regression designs, the approach of this paper is different in that it uses Bayesian decision theory rather than orthodox (classical) methods. Lindley (1968) considered the prediction problem given the results of the regression experiment, and we shall be using some of his results in this paper.

SUMMARY Bayesian decision theory is used to obtain optimal designed and random regression experiments. If a choice is possible, it is shown how to decide between them. This is done if the data obtained are to be analyzed in order that a future value of the dependent variable can be predicted using the best subset of the independent variables.

Papers that have been written on the use of the rank order statistic for the two-sample problem under the Lehmann alternatives have focused on the use of rank tests to determine whether there is a difference in the two populations. In particular, the test proposed by Savage (1956) is the rank test that is usually suggested to be used in this situation. In this paper a Bayesian analysis of the problem based on the rank order statistic is presented. Suppose that a random sample of m observations is obtained from distribution 1 and an independent random sample of n observations is obtained from distribution 2, and N = m+n. The rank order statistic will be denoted by z = (z1, ...,ZN), where zi is 0 if the ith smallest observation is from distribution 1 and zi is 1 otherwise (i -1, ..., N). Let F(x) and G(x) be the distribution functions of distributions 1 and 2, respectively, both assumed to be continuous. If G(x) is the Lehmann alternative 1{1 F(x)}k(k > 0), which holds, for example, if the population distributions are both exponential or both Weibull, then the rank order probability is

One hundred fifty-six women with axillary node-negative breast cancer and primary tumors less than or equal to 5 cm in diameter (T1N0 or T2N0) were treated with a brief course of postoperative adjuvant chemotherapy consisting of doxorubicin and cyclophosphamide. Treatment was well tolerated and toxicity was minimal. With a median follow-up time of 58 months, there has been 1 relapse among 58 patients with T1 primary lesions and 15 relapses among 98 patients with T2 primary tumors. When compared with a matched historical control group receiving surgery alone, significant improvement was apparent in disease-free survival among the patients who received adjuvant chemotherapy. Prospective controlled trials are needed if we are to confirm this favorable experience with adjuvant chemotherapy in the treatment of women with node-negative breast cancer.

The development of the sarcocyst of Sarcocystis rauschorum in its intermediate host was studied. Lemmings were orally administered sporocysts of S. rauschorum obtained from snowy owls (Nyctea scandiaca). Beginning at 9 days postinoculation (DPI) and at various intervals to 84 DPI, skeletal muscle tissue taken from the infected lemmings was examined by electron microscopy. At 9 DPI the sarcocysts contained few metrocytes and the cyst wall was flat. The metrocytes underwent endodyogeny, and within a few days the cyst wall of the rapidly growing sarcocyst developed numerous tubulovesicular invaginations into the electron-dense layer, and the wall had a few irregular infoldings. By 21 DPI, banana-shaped bradyzoites appeared, and by 84 DPI the mature cysts were filled with bradyzoites in groups subdivided by septa and by deep infoldings of the cyst wall. The fine structure of the wall remained simple throughout maturation, with no conspicuous invagination or protrusion. The sarcocyst produced in response to S. rauschorum is unlike those from many species of Sarcocystis, which have complex walls that change markedly as the sarcocysts mature; however, its simple appearance is similar to other species that have rodents as intermediate hosts and raptorial birds as definitive hosts.

In this paper we update the results of a series of clinical trials of chemotherapy and radiotherapy for patients with early stage breast cancer seen and treated at the University of Arizona Cancer Center between 1974 and 1984. Preliminary results have been presented at earlier conferences on the adjuvant therapy of cancer [1–7]. The basic chemotherapy employed in these adjuvant trials consisted of the two-drug combination of doxorubicin (adriamycin) and cyclophosphamide (AC) administered at 3-week intervals. These agents were picked for study because of their efficacy in treating patients with advanced breast cancer, the subsequent evidence of drug synergism from animal tumor systems, and for their ease of administration in an outpatient setting. This report includes data based on an analysis made in October, 1983.

Malignant neoplasms were present in 39/66 Dicrostonyx groenlandicus of varying ages, examined from a laboratory colony. The presence of multiple neoplasms in some resulted in an overall average of 1.15 tumors/affected animal. Gastric squamous papillomas were present in nine, and locally invasive or metastatic gastric squamous cell carcinomas in a further 36 animals. Three mammary adenocarcinomas, one pancreatic islet cell tumor, one probable pancreatic adenocarcinoma and one adrenal cortical adenoma were also seen in lemmings with gastric tumors. Two others had mammary adenocarcinoma alone, and one animal had bilateral Harderian gland adenocarcinoma. Lesions resembling glomerulonephrosis of rats were seen in 23/51 animals whose kidneys were examined. These findings were not considered artefacts of captivity since concurrent gastric squamous cell carcinoma, mammary adenocarcinoma and glomerulonephrosis were present in the single animal examined directly from the wild at Eskimo Point, Northwest Territories.

Peripheral blood counts and drug dosages were studied in 68 patients with advanced non-Hodgkin lymphoma, who were assigned at random to receive either chemoimmunotherapy or chemotherapy, in order to determine if Bacillus Calmette Guerin (BCG) immunotherapy lessened myelosuppression or improved drug tolerance. Thirty-seven patients received CHOP (cyclophosphamide, adriamycin, vincristine, and prednisone) plus BCG, and 31 patients received CHOP plus low-dose bleomycin. These two treatment groups, similar for important clinical and prognostic factors, were evaluated at courses one, four, and eight of remission induction, by comparison of the following parameters: hemoglobins, hematocrits, platelet counts, white blood-cell counts (including absolute neutrophil, lymphocyte, and monocyte counts), and dosages of cyclophosphamide and adriamycin. With successive treatment both groups experienced statistically significant progressive declines in the mean values of hemoglobin, hematocrit, total white blood-cell count, absolute granulocyte and lymphocyte counts, and dosages of adriamycin and cyclophosphamide. These effects of treatment were comparable except for a slightly greater degree of mild anemia in the CHOP-BCG group (mean hemoglobin ± SEM = 11.7 ± 0.3 gm/dl at course eight compared to 12.7 ± 0.3 gm/dl for the CHOP-bleomycin group, p = 0.04). Thus, BCG appears to play no beneficial role in lessening myelosuppression resulting from chemotherapy or improving drug tolerance.

A Bayesian formulation of the problem of analysing data from a curvilinear regression of $y$ on $x_1, x_2, \cdots, x_r$ in order to predict a future value of $y$ is considered. The problem is to obtain a criterion to decide which is the best subset of $x_1, x_2, \cdots, x_r$ to perform this prediction. Under very strict assumptions the criterion obtained is shown to use the same statistic as the orthodox (least squares) approach.

Continuous intravenous or subcutaneous infusion of drugs is increasingly common in many specialties. This increase reflects a better understanding of drug pharmacokinetics and a greater availability of infusion controllers and syringe pumps. Although local limitations on many nurses prevent them from administering drugs intravenously by "bolus" injections, they are allowed to control infusions. The use of infusions has induced errors resulting from their arithmetical complexity, which has even stimulated the writing of computer programs. Many factors contribute to the complexity and we think that they have three main origins. The first is the different requirements of the medical and nursing staff concerned in the administration of the infusion. Medical staff wish to refer to the dose of drug per kilogram of body weight per unit of time. Only by using this standard can effect be related to dose and similar drug regimens compared. This value is, however, lost from sight by the need of the nursing staff to have a prescription based on the volume flow rate of the infusion, as this is what they actually control. Secondly, the relation between the drug dosage rate and the volume is obfuscated by a bewildering array of expressions of concentration?percentages (vol/vol or wt/vol), parts per thousand, parts per mule (%o), mass or moles per 100 ml, per litre, or per ml. When a solution for infusion is made up on the ward probably the concentration will be expressed by describing the method of preparation?for example, "0*15 mg of fentanyl in 30 ml of saline." Apart from the ambiguity (was the solution made up to or with 30 ml saline?), a knowledge of the concentration of the original drug solution and a degree of arithmetic agility are required to determine the nature and strength of the infused solution and hence the dose being given to the patient. Manufacturers often suggest a suitable dilution method without stating the concentration obtained, or suggest an infusion rate without giving the drug dose rate. In such obscurities lies a potential source of serious error. Thirdly, infusions perforce require the intake of liquid which may contribute appreciably to the patient's total intake, and this is especially true in smaller patients and when several infusions are given simultaneously. Although no single system can be universally applicable, it is desirable to have uniformity of expression of all the values required to prescribe an infusion. Thus we need (a) a unit of time (minute/hour/day) and (b) a unit of volume (ml/100 ml/litre).