A statistical model for predicting response of breast cancer patients to cytotoxic chemotherapy.

A binary logistic model is used for predicting response to cytotoxic chemotherapy for a breast cancer patient on the basis of her tumor enzyme activity profile. The enzymes used in the model are lactate dehydrogenase, nicotinamide adenine dinucleotide phosphate-isocitrate dehydrogenase, and phosphoglucomutase, all of which were measured on primary tumor specimens from each patient. The statistical model provides an estimate of the probability that an individual will respond to treatment. Chemotherapeutic treatment consisting of combination cytotoxic drugs and subsequent evaluation of patient response followed cooperative group protocol guidelines, including outside review to confirm the patient evaluation. The model based on this study, which represents 5 years of patient follow-up, correctly predicts clinical outcome in 32 of the 37 cases available.

Fundamental dilemmas of the randomized clinical trial process: results of a survey of the 1,737 Eastern Cooperative Oncology Group investigators.

PURPOSE: We studied oncologists' attitudes and behavior with regard to their participation in randomized clinical trials. METHODS: We surveyed the 1,737 physician members of the Eastern Cooperative Oncology Group (ECOG) using the Physician Orientation Profile (POP), a self-administered mailed questionnaire. A response rate of 86% was achieved (1,485 of 1,737); each physician's actual patient accrual was recorded. RESULTS: All respondents indicated that they had a systematic pattern of patient preselection for entry onto trials beyond the formal inclusion/exclusion trial criteria. Eighty-nine percent stated that improving patient quality of life rather than prolonging survival was more personally satisfying. Sixty-two percent did not enter a single patient during the 12-month period following the survey, while 10% entered 80% of all patients during that time. Physicians overestimated their accrual rate by a factor of 6. Eighty-three percent defined randomization and adherence to trial protocol as a serious challenge to their ability to make individualized treatment decisions. CONCLUSION: This study raises questions regarding the following: (1) the perceived generalizability of trial findings, (2) the role of end points other than survival for clinical trials, (3) the consequences of physician overestimation of patient accrual, and (4) the impact of randomized trials on the behavior of clinicians. Further investigation into these critical issues will provide meaningful recommendations to enhance the future design, implementation, and conduct of randomized clinical trials in cancer.

Recombinant tissue-type plasminogen activator versus a novel dosing regimen of urokinase in acute pulmonary embolism: a randomized controlled multicenter trial.

Thrombolysis of acute pulmonary embolism can be accomplished more rapidly and safely with 100 mg of recombinant human tissue-type plasminogen activator (rt-PA) (Activase) than with a conventional dose of urokinase (Abbokinase) given as a 4,400-U/kg bolus dose, followed by 4,400 U/kg per h for 24 h. To determine the effects of a more concentrated urokinase dose administered over a shorter time course, this trial enrolled 90 patients with baseline perfusion lung scans and angiographically documented pulmonary embolism. They were randomized to receive either 100 mg/2 h of rt-PA or a novel dosing regimen of urokinase: 3 million U/2 h with the initial 1 million U given as a bolus injection over 10 min. Both drugs were delivered through a peripheral vein. To assess efficacy after initiation of therapy, repeat pulmonary angiograms at 2 h were performed in 87 patients and then graded in a blinded manner by a panel of six investigators. Of the 42 patients allocated to rt-PA therapy, 79% showed angiographic improvement at 2 h, compared with 67% of the 45 patients randomized to urokinase therapy (95% confidence interval for the difference in these proportions [rt-PA minus urokinase] is -6.6% to 30.4%; p = 0.11). The mean change in perfusion lung scans between baseline and 24 h was similar for both treatments. Three patients (two treated with rt-PA and one with urokinase) had an intracranial hemorrhage, which was fatal in one. The results indicate that a 2-h regimen of rt-PA and a new dosing regimen of urokinase exhibit similar efficacy and safety for treatment of acute pulmonary embolism.

Lung scan evaluation of thrombolytic therapy for pulmonary embolism.

UNLABELLED: Data from three trials of thrombolytic therapy for pulmonary embolism (PE) were combined to assess the utility of perfusion lung scan defect scoring in predicting the response to thrombolytic therapy. METHODS: Pre- and post-therapy lung scans and duration of symptoms were available for a total of 221 patients, 167 were treated with various thrombolytic regimes and 54 were treated with heparin alone. RESULTS: Improvement in the lung scan defect score was correlated with larger initial defect score (r = 0.53), segmental appearance (r = 0.31) and shorter duration of symptoms (r = 0.20). There was no significant residual correlation between improvement and segmental appearance in a multiple regression analysis after accounting for initial defect score and duration of symptoms. Two lung scan scoring methods (segmental and anterior-posterior method) provided similar results with low interobserver variability (r = 0.90 for both methods). CONCLUSION: This study indicates that the baseline perfusion lung scan defect severity helps to predict the response to thrombolytic therapy.

Efficacy and safety of repeated boluses of urokinase in the treatment of deep venous thrombosis.

The only Food and Drug Administration-approved thrombolytic regimen for treatment of deep venous thrombosis (DVT) is a 24- to 72-hour continuous infusion of intravenous streptokinase. This approach to DVT thrombolysis is not entirely satisfactory because of the bleeding complications that may accompany this therapy. In the current study, we treated 27 patients with DVT with a novel dosing regimen of urokinase: 1,000,000 U administered as a 10-minute bolus, with a total of 3 boluses given over approximately 24 hours. Patients were given heparin overnight between bolus urokinase doses. Efficacy was assessed by comparing baseline and prehospital discharge vascular imaging studies, which constituted either venous ultrasound or contrast venography. A vascular-imaging panel of physicians, unaware of the sequence of paired studies, found that 14 patients (52%) had clot lysis (6 slight, 6 moderate and 2 marked), 9 (33%) had no change, and 4 (15%) had more extensive thrombosis after treatment (1 slight, 2 moderate and 1 marked). There were no bleeding complications. At 48 hours after starting urokinase, mean plasma fibrinogen levels had decreased 61% from baseline, and the mean bleeding time had increased 28% from baseline (but remained within the normal range). Because of the promising efficacy and safety that were found in this case series, it is concluded that further testing of bolus urokinase is warranted against anticoagulation alone.

Heparin inhibits bovine testicular hyaluronidase activity in myocardium of dogs with coronary artery occlusion.

Bovine testicular hyaluronidase (BTH) reduces experimental myocardial infarct size and ameliorates electrocardiographic signs of ischemia. This study was done to determine if heparin, an in vitro inhibitor of hyaluronidase activity, blocks the action of BTH in the myocardium of dogs after coronary artery occlusion. BTH was administered intravenously as 5,000 NF units/kg at 0.5 and 2.5 hours after coronary occlusion. Heparin was administered intravenously as a 150-unit/kg loading dose, followed by 10 units/kg per hour i.v., beginning 15 minutes before coronary occlusion. The area of myocardial ischemia at risk was assessed by a radiolabeled microsphere technique; the area that developed necrosis was assessed by a histochemical technique. In vivo activity of BTH was assessed by a colorimetric analysis of the BTH substrate, i.e., hyaluronic acid (HA), extracted from myocardial tissue. For biochemical analysis of HA, the heart was divided into anterior myocardium, which included ischemic tissue and posterior nonischemic myocardium. The myocardial HA content of dogs treated with BTH plus heparin (anterior, 3.44 +/- 0.40 micrograms HA/mg protein; posterior, 3.69 +/- 0.33 micrograms HA/mg protein) was not significantly different from control (anterior, 3.61 +/- 0.29 micrograms HA/mg protein; posterior, 3.55 +/- 0.23 micrograms HA/mg protein). In contrast, BTH lowered myocardial HA content (anterior, 2.16 +/- 0.21 micrograms HA/mg protein; posterior, 2.08 +/- 0.14 micrograms HA/mg protein) compared with either BTH plus heparin or control groups in both anterior myocardium (p = 0.006) and posterior myocardium (p = 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)

Two trials of reduced bolus alteplase in the treatment of pulmonary embolism. An overview.

The Bolus Alteplase Pulmonary Embolism (BAPE) Group and a consortium of French investigators utilized essentially the same investigational protocol to test reduced dose bolus alteplase vs full dose 100 mg/2 h alteplase in the treatment of pulmonary embolism (PE). The principal hypothesis was that reduced dose bolus alteplase (n = 96) would result in fewer bleeding complications than full dose 100 mg of 2 h alteplase (n = 44) administered as a continuous infusion to hemodynamically stable patients with PE. To provide data on bolus alteplase's safety profile in a larger sample size than would have been feasible in either trial alone, we present an overview of the BAPE and French trials. There were no differences between the reduced dose bolus and full dose 2 h rt-PA groups with respect to bleeding complications. Therefore, the principal hypothesis of these two randomized controlled trials could not be confirmed. Efficacy was similar in the two treatment groups. Interpretation of the results will vary because the increased convenience and cost savings from using a reduced dose of bolus alteplase may be offset by a higher mortality rate. However, a trial that compared the mortality rates of the two treatment regimens would have required more than 800 patients.

Correlates of patient satisfaction with mental health services in a health maintenance organization.

To assess the level and determinants of patient satisfaction with prepaid mental health services, the authors surveyed 124 adult patients who five years earlier had been seen at least once by a mental health professional. Of the 44 percent who returned questionnaires, approximately one third reported satisfaction, one third dissatisfaction, and one third were unsure. Most patients sought help for chronic situational problems and tended to use the services repeatedly over time. Satisfaction was most strongly correlated with the patients' perceptions of their rapport with the provider and slightly less so with their perceptions of the provider's technical competence. Patient characteristics, diagnoses according to the Diagnostic and Statistical Manual of Mental Disorders (DSM 3), and systems issues were of no predictive value. Patients who presented with symptomatic distress not clearly linked to circumstances were likely to be dissatisfied with treatment.

Characteristics of diet among a culturally diverse group of low-income pregnant women.

This study reports on diet quality and variety in food selection among a culturally diverse group of 335 low-income pregnant Massachusetts women. The Index for Nutritional Quality (INQ), which is the observed nutrient density divided by the recommended nutrient density, was estimated for seven nutrients from data reported on food frequency questionnaires and diet recalls. Mean INQs for protein and vitamin C were above the recommended level of 1.0, whereas INQ for iron was 0.6 or less. Mean INQ for calcium was 1.2 for whites and between 0.9 and 1.0 for non-whites. When comparisons were made among ethnic groups by site of prenatal care, significant differences in INQ were found for all nutrients except iron and vitamin B-6. WIC participants had higher diet quality for protein, calcium, iron, and vitamin B-6 than did non-participants. Employed women had higher protein and zinc INQs than did unemployed women, and non-smokers had higher iron and vitamin B-6 INQs than did smokers. Variety was not significantly correlated with diet quality but differed (p less than .001) among site-specific ethnic groups. We conclude that there is a need for investigation of factors influencing dietary practices that are associated with sites of prenatal care.

Age-specific patterns of diagnosis of acute otitis media.

A retrospective review of clinical experience with the diagnosis of acute otitis media (AOM) in a large pediatric population (age, 0-12 years old) reveals two age-specific patterns of diagnosis: one in preschool children and one in schoolage children. In the population as a whole, AOM is diagnosed most frequently in the first 5 years of life; rates of diagnosis vary with gender (male: female ratio = 1.10) and season (fall/winter peak). In a linear regression model, age, gender, season, and interactions among these account for 52 percent of the variability in diagnostic rates. In children less than 5 years of age, rates of diagnosis vary strikingly with gender and season; the regression model accounts for 55 percent of observed variability in diagnostic rates. In schoolage children (5-11 years old), however, males and females are equally likely to be diagnosed to have AOM, and seasonal variability is blunted; among these children, the regression model predicts only 22 percent of diagnostic rate variability. These findings buttress existing knowledge of the epidemiology of AOM and have important implications for clinical practice, health planning, and pediatric research.

Quality-of-life-adjusted survival for comparing cancer treatments. A commentary on TWiST and Q-TWiST.

In chronic disease situations where treatment comparisons favor no particular therapy, or where definitive outcome requires a considerable follow-up period, it is useful to have additional and perhaps intermediate endpoints of relevant clinical significance to compare treatments. One such endpoint is Time Without Symptoms and Toxicity (TWiST) which, together with Q-TWiST, attempts to address the quality of life of patients receiving the competing regimens. This paper provides a commentary on these techniques with an emphasis on the problems inherent in implementing Q-TWiST, a measure that attempts to incorporate patient value preferences into TWiST. It is argued that while Q-TWiST is intuitively appealing in the clinical setting, there are formidable design and psychometric hurdles that must be overcome to fully operationalize the concept.

Clinical trials in psychosocial medicine: methodologic and statistical considerations. Part II. Assessing reliability with equal-appearing interval scales.

The usual statistical procedures for assessing reliability among several raters do not apply to scales of the equal-appearing interval type. This is so primarily because the set of possible responses on such scales, namely the nonnegative integers 0,1,...(M-1), contains so few points that normal distribution theory cannot be invoked. Traditional analyses of such data, which include the one-way analysis of variance intraclass correlation, produce a reliability coefficient based on an assumed normally distributed response variable, and must therefore be interpreted with great caution. This paper discusses a statistical approach which exploits the discrete nature of the response variable and hence is more appropriate for such data.

Capillary lengths, anastomoses, and estimated capillary transit times in skeletal muscle.

Total capillary length, capillary segment length, and number of anastomoses per capillary were measured in rat gracilis muscle at rest and after 2 min of phasic contraction. Mean values of the foregoing variables at rest (+/-SD) were, respectively, 1,012 micronm +/- 484, 409 micron +/- 274, and 0.83 +/- 1.09. Total capillary lengths are well described by the gamma distribution, number of anastomoses by the negative binomial distribution, and segment length by the Weibull distribution. Contraction has no significant effect on the means or the frequency distributions, indicating that: 1) pressure gradients between adjacent capillaries are small, and 2) intercapillary anastomoses do not improve flow distribution in exercise. Erythrocyte velocities observed in resting muscle (Burton, K. S., and P. C. Johnson. Am J. Physiol. 223: 517-524, 1972) were shown to be adequately characterized by the gamma distribution. From these velocities and the observed distribution of path lengths, we computed an estimated distribution of capillary transit times. Mean transit time was 4.29 s. The median was 2.45 s, and 11% of values exceeded 8 s. The range was 90 ms-43 s. This heterogeneity of transit times should profoundly affect calculations of O2 transport and the shape of indicator dilution curves.