Strain-level diversity of symbiont communities between individuals and populations of a bioluminescent fish

The bioluminescent symbiosis involving the urchin cardinalfish, Siphamia tubifer, and Photobacterium mandapamensis, a luminous member of the Vibrionaceae, is highly specific compared to other bioluminescent fish-bacteria associations. Despite this high degree of specificity, patterns of genetic diversity have been observed for the symbionts from hosts sampled over relatively small spatial scales. We characterized and compared sub-species, strain-level symbiont diversity within and between S. tubifer hosts sampled from the Philippines and Japan using PCR fingerprinting. We then carried out whole genome sequencing of the unique symbiont genotypes identified to characterize the genetic diversity of the symbiont community and the symbiont pangenome. We determined that an individual light organ contains six symbiont genotypes on average, but varied between 1-13. Additionally, we found that there were few genotypes shared between hosts from the same location. A phylogenetic analysis of the unique symbiont strains indicated location-specific clades, suggesting some genetic differentiation in the symbionts between host populations. We also identified symbiont genes that were variable between strains, including luxF, a member of the lux operon, which is responsible for light production. We quantified the light emission and growth rate of two strains missing luxF along with the other strains isolated from the same light organs and determined that strains lacking luxF were dimmer but grew faster than most of the other strains, suggesting a potential metabolic trade-off. This study highlights the importance of strain-level diversity in microbial associations and provides new insight into the underlying genetic architecture of intraspecific symbiont communities within a host.

Chromosome-level genome assembly of the bioluminescent cardinalfish Siphamia tubifer, an emerging model for symbiosis research

The bioluminescent symbiosis involving the sea urchin cardinalfish Siphamia tubifer and the luminous bacterium Photobacterium mandapamensis is an emerging vertebrate model for the study of microbial symbiosis. However, little genetic data are available for the host, limiting the scope of research that can be implemented with this association. We present a chromosome-level genome assembly for S. tubifer using a combination of PacBio HiFi sequencing and Hi–C technologies. The final assembly was 1.2 Gb distributed on 23 chromosomes and contained 32,365 protein coding genes with a BUSCO score of 99%. A comparison of the S. tubifer genome to that of another nonluminous species of cardinalfish revealed a high degree of synteny, whereas a comparison to a more distant relative in the sister order Gobiiformes revealed the fusion of two chromosomes in the cardinalfish genomes. The complete mitogenome of S. tubifer was also assembled, and an inversion in the vertebrate WANCY tRNA genes as well as heteroplasmy in the length of the control region were discovered. A phylogenetic analysis based on whole the mitochondrial genome indicated that S. tubifer is divergent from the rest of the cardinalfish family, highlighting the potential role of the bioluminescent symbiosis in the initial divergence of Siphamia. This high-quality reference genome will provide novel opportunities for the bioluminescent S. tubifer–P. mandapamensis association to be used as a model for symbiosis research.

Environmental modifiable risk factors for multiple sclerosis: Report from the 2016 ECTRIMS focused workshop

Multiple sclerosis (MS) is an inflammatory and neurodegenerative demyelinating disease of the central nervous system (CNS), most likely autoimmune in origin, usually beginning in early adulthood. The aetiology of the disease is not well understood; it is viewed currently as a multifactorial disease which results from complex interactions between genetic predisposition and environmental factors, of which a few are potentially modifiable. Improving our understanding of these factors can lead to new and more effective approaches to patient counselling and, possibly, prevention and management of the disease. The 2016 focused workshop of the European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS) addressed the topic of environmental, modifiable risk factors for MS, gathering experts from around the world, to collate experimental and clinical research into environmental factors that have been associated with the disease onset and, in a few cases, disease activity and progression. A number of factors, including infections, vitamin D deficiency, diet and lifestyle factors, stress and comorbidities, were discussed. The meeting provided a forum to analyse available evidence, to identify inconsistencies and gaps in current knowledge and to suggest avenues for future research.

Life history of the symbiotically luminous cardinalfish Siphamia tubifer (Perciformes: Apogonidae)

Characteristics of the life history of the coral reef-dwelling cardinalfish Siphamia tubifer, from Okinawa, Japan, were defined. A paternal mouthbrooder, S. tubifer, is unusual in forming a bioluminescent symbiosis with Photobacterium mandapamensis. The examined S. tubifer (n = 1273) ranged in size from 9·5 to 43·5 mm standard length (LS ), and the minimum size at sexual maturity was 22 mm LS . The number of S. tubifer associated during the day among the spines of host urchins was 22·9 ± 16·1 (mean ± s.d.; Diadema setosum) and 3·6 ± 3·2 (Echinothrix calamaris). Diet consisted primarily of crustacean zooplankton. Batch fecundity (number of eggs; FB ) was related to LS by the equations: males (fertilized eggs) FB  = 27·5LS  - 189·46; females (eggs) FB  = 31·3LS  - 392·63. Individual mass (M; g) as a function of LS was described by the equation: M=9·74×10-5LS2·68. Growth, determined from otolith microstructure analysis, was described with the von Bertalanffy growth function with the following coefficients: L∞  = 40·8 mm LS , K = 0·026 day(-1) and t0  = 23·25 days. Planktonic larval duration was estimated to be 30 days. The age of the oldest examined individual was 240 days. The light organ of S. tubifer, which harbours the symbiotic population of P. mandapamensis, increased linearly in diameter as S. tubifer LS increased, and the bacterial population increased logarithmically with S. tubifer LS . These characteristics indicate that once settled, S. tubifer grows quickly, reproduces early and typically survives much less than 1 year in Okinawa. These characteristics are generally similar to other small reef fishes but they indicate that S. tubifer experiences higher mortality.

Symbiosis initiation in the bacterially luminous sea urchin cardinalfish Siphamia versicolor

To determine how each new generation of the sea urchin cardinalfish Siphamia versicolor acquires the symbiotic luminous bacterium Photobacterium mandapamensis, and when in its development the S. versicolor initiates the symbiosis, procedures were established for rearing S. versicolor larvae in an aposymbiotic state. Under the conditions provided, larvae survived and developed for 28 days after their release from the mouths of males. Notochord flexion began at 8 days post release (dpr). By 28 dpr, squamation was evident and the caudal complex was complete. The light organ remained free of bacteria but increased in size and complexity during development of the larvae. Thus, aposymbiotic larvae of the fish can survive and develop for extended periods, major components of the luminescence system develop in the absence of the bacteria and the bacteria are not acquired directly from a parent, via the egg or during mouth brooding. Presentation of the symbiotic bacteria to aposymbiotic larvae at 8-10 dpr, but not earlier, led to initiation of the symbiosis. Upon colonization of the light organ, the bacterial population increased rapidly and cells forming the light-organ chambers exhibited a differentiated appearance. Therefore, the light organ apparently first becomes receptive to colonization after 1 week post-release development, the symbiosis is initiated by bacteria acquired from the environment and bacterial colonization induces morphological changes in the nascent light organ. The abilities to culture larvae of S. versicolor for extended periods and to initiate the symbiosis in aposymbiotic larvae are key steps in establishing the experimental tractability of this highly specific vertebrate and microbe mutualism.

Baseline differences in A1C explain apparent differences in efficacy of sitagliptin, rosiglitazone and pioglitazone

AIM

Published studies of patients treated with rosiglitazone or pioglitazone have reported greater reductions in HbA1c (A1C) than studies of patients treated with sitagliptin. However, studies of thiazolidinediones tended to enroll patients with higher baseline A1C levels. This meta-analysis investigates the relationship between baseline A1C and perceived efficacy of treatment.

METHODS

This report describes a Bayesian random effects analysis of 23 published studies. We constructed a random effects model including a factor adjusting for between-study differences in baseline A1C levels.

RESULTS

The random effects model correctly predicts post-treatment A1C levels from baseline A1C within a 95% confidence interval (CI) for each of the 23 studies included in the meta-analysis. After applying the model to adjust for differences in baseline A1C, we found that the difference in efficacy between rosiglitazone and sitagliptin was not significantly different from zero (0.12; 95% CI -0.09 to 0.34). Similarly, no significant differences are observed between the effects of pioglitazone and sitagliptin (0.01; 95% CI -0.21 to 0.22) or between rosiglitazone and pioglitazone (0.11; 95% CI -0.37 to 0.146). When baseline values are omitted from the Bayesian model, the findings suggest that rosiglitazone is superior to pioglitazone or sitagliptin.

CONCLUSIONS

These results illustrate the necessity for careful application of appropriate methodology when comparing results of different studies. When between-study differences in treatment effects are adjusted for baseline differences, then the findings suggest that none of the treatments has an effect that is superior to any of the other treatments.

An evaluation of computer-aided disproportionality analysis for post-marketing signal detection

To understand the value of computer-aided disproportionality analysis (DA) in relation to current pharmacovigilance signal detection methods, four products were retrospectively evaluated by applying an empirical Bayes method to Merck's post-marketing safety database. Findings were compared with the prior detection of labeled post-marketing adverse events. Disproportionality ratios (empirical Bayes geometric mean lower 95% bounds for the posterior distribution (EBGM05)) were generated for product-event pairs. Overall (1993-2004 data, EBGM05> or =2, individual terms) results of signal detection using DA compared to standard methods were sensitivity, 31.1%; specificity, 95.3%; and positive predictive value, 19.9%. Using groupings of synonymous labeled terms, sensitivity improved (40.9%). More of the adverse events detected by both methods were detected earlier using DA and grouped (versus individual) terms. With 1939-2004 data, diagnostic properties were similar to those from 1993 to 2004. DA methods using Merck's safety database demonstrate sufficient sensitivity and specificity to be considered for use as an adjunct to conventional signal detection methods.

Sample size re-estimation: recent developments and practical considerations

Interim findings of a clinical trial often will be useful for increasing the sample size if necessary to provide the required power against the null hypothesis when the alternative hypothesis is true. Strategies for carrying out the interim examination that have been described over the past several years include "internal pilot studies", blinded interim sample size adjustment and conditional power. Simulation studies show that the alternative methods generally control the type I error rate satisfactorily, although the power properties are more variable. The important issues associated with sample size re-estimation are strategic, not numeric. Clearly expressed regulatory preferences suggest that methods not requiring unblinding the data before completion of the trial would be most appropriate. Extending a trial has its risks. The investigators/patients enrolled later in the course of a trial are not necessarily the same as those recruited/entered early. Re-activating the enrollment process may be sufficiently complicated and expensive to justify enrolling more investigators/patients at the outset. Since sample size re-estimation adjusts the sample size on the basis of variability while efficacy interim analysis adjusts the sample size based on the basis of estimated effect size, both principles can be used in the same trial. Sample size re-estimation may not be advisable for trials involving extended follow-up of individual patients or, more generally, when the follow-up time is long relative to the recruitment time. In such cases, it may be better to estimate the sample size conservatively and introduce an interim efficacy evaluation.

A practical approach for evaluating population and individual bioequivalence

Pharmacokinetic measurements provided by subjects to each of two formulations of a drug have a joint distribution that can be characterized by parameters reflecting scale and correlation as well as location. The bioavailability of the formulations can be expressed in terms of the means of the marginal distributions, their means and variances, or the marginal means and variances and the joint correlation. These expressions correspond, respectively, to 'average', 'population', and 'individual' bioequivalence when the joint distribution of the measurements is bivariate normal. Current proposals for assessing the degree of bioequivalence of two formulations are based on statistics that are composites of variance components and squares of expected mean differences from a mixed linear model. There are technical and practical issues associated with these proposals, particularly that they require more complicated designs than the familiar 2x2 cross-over. This paper describes an alternative approach that can be applied with standard 2x2 cross-over designs, and that provides evaluations of population and individual bioequivalence that should be adequate for all practical clinical purposes. The approach is based on easily computed correlation and regression coefficients whose statistical properties under normality are well known and for which non-parametric and robust alternatives exist when normality cannot be assumed. The approach yields conclusions consistent with those obtained by the current proposals when applied to data sets supplied by the FDA. In the cases where the conclusions do not match, the new approach appears to be more consistent with the data.

PhRMA perspective on population and individual bioequivalence

The Food and Drug Administration (FDA) issued a second-draft guidance in August 1999 on the subject of in vivo bioequivalence, which is based on the concepts of individual and population bioequivalence (IBE and PBE, respectively). The intention of this guidance is to replace the 1992 guidance that requires that in vivo bioequivalence be demonstrated by average bioequivalence (ABE). Although the concepts of population and individual bioequivalence are intuitively reasonable, a detailed review of the literature has not uncovered clinical evidence to justify the additional burden to the innovator and generic companies as well as the consumer that the new guidelines would impose. The criteria for bioequivalence described in the draft guidance employ aggregate statistics that combine information about differences in bioavailability between formulation means and differences in bioavailability variation of formulations between and within subjects. The purely technical aspects of the statistical approach are reasonably sound. However, PhRMA believes that important operational issues remain that need to be resolved before any changes to current practice are implemented. PhRMA believes that the ideals of prescribability and switchability are intuitively reasonable, but it is uncertain of the extent to which the proposed guidance can achieve these goals. It is not clear whether the attainment of such goals is necessary in the evaluation of bioequivalence given the role this plays in drug development, and the lack of clinical evidence argues against a pressing need to change current practice. PhRMA is concerned that the trade-off offered by the aggregate criteria may ultimately represent more harm than good to the public interest. PhRMA recommends more rigorous evaluation of methods based on two-way crossover designs before moving to methods that require more complex designs. One such method is identified herein and contains procedures for estimating prescribability and switchability. The possibility of a phase-in or trial period to collect replicate crossover data to further evaluate IBE and PBE and possibly allow market access based on these criteria as they are being evaluated has been proposed. PhRMA believes this is unprecedented and will offer little additional information beyond that which can be obtained by simulation or has already been collected by the FDA. Simulation studies have the advantage of allowing evaluation of the sensitivity of various procedures to represent the data patterns as created within the simulation. Operating characteristics by which proposed criteria can be adequately judged have not yet been defined. The limitations of ABE for highly variable drugs and narrow therapeutic drugs are well appreciated and may be addressed by means other than a wholesale change in the current criteria.

Serum prostate-specific antigen as a predictor of prostate volume in men with benign prostatic hyperplasia

OBJECTIVES

To assess the utility of prostate-specific antigen (PSA) as a predictor of prostate volume by characterizing the relationship between prostate volume and serum PSA in men with symptomatic benign prostatic hyperplasia (BPH) and no evidence of prostate cancer, stratified by decade of life.

METHODS

Placebo-controlled multicenter trials in patients with BPH and a safety study in normal young men provided baseline measurements of serum PSA and prostate volume. The analyses included patients with a baseline prostate volume measured by either transrectal ultrasound (TRUS) or magnetic resonance imaging and baseline serum PSA. A common central laboratory was used for all but one of the individual studies; both laboratories used the Hybritech method. Patients 80 years of age or older were excluded. Patients with a baseline serum PSA greater than 10 ng/mL were excluded to reduce the likelihood of including occult prostate cancer cases. The patients in the BPH trials were screened at baseline by digital rectal examination (DRE) and serum PSA. Those with suspicious findings underwent TRUS-guided biopsy; only patients with negative biopsies are included in these analyses.

RESULTS

The analyses included 4627 patients, 4448 from the BPH trials and 179 from the safety study. The men in the BPH trials were older (mean age+SE, 63.7+0.10 years) than the men in the safety study (mean age + SE, 30.8+/-0.43), had larger prostates (mean volume+/-SE, 43.7+/-0.38 mL versus 26.3+/-0.49 mL in the safety study), and had higher serum PSA values (mean+/-SE, 2.6+/-0.03 ng/mL versus 0.7+/-0.39 ng/mL in the safety study). The relationship between prostate volume and serum PSA was evaluated using only the BPH trial data. Prostate volume and serum PSA have an age-dependent log-linear relationship (ie, their logarithms are linearly related, and the parameters of the relationship depend on age). Older men tend to have a steeper rate of increase in prostate volume with increasing serum PSA (P < 0.00 for differences between slopes), and there was a slight tendency for PSA density to increase with age. Receiver operating characteristic (ROC) curves were constructed to evaluate the ability of serum PSA to predict threshold prostate sizes in men with BPH. The ROC curve analyses revealed that PSA had good predictive value for assessing prostate volume, with areas under the curve ranging from 0.76 to 0.78 for various prostate volume cutoff points (30, 40, and 50 mL). Conclusions. Prostate volume is strongly related to serum PSA in men with BPH and no evidence of prostate cancer, and the relationship depends on age. Since treatment outcome or risk of long-term complications depend on baseline prostate volume, serum PSA can estimate the degree of prostate enlargement sufficiently accurately to be useful for therapeutic decision making. To achieve a specificity of 70% while maintaining a sensitivity between 65% and 70%, approximate age-specific criteria for detecting men with prostate glands exceeding 40 mL are PSA > 1.6 ng/mL, >2.0 ng/mL, and >2.3 ng/mL for men with BPH in their 50s, 60s, and 70s, respectively.

Multi-centre trial analysis revisited

Analyses of multi-centre trials must consider the effects of the individual centres and the possibility of non-constancy of treatment effect differences among centres. This usually means an ANOVA with terms for centres, treatments, and centre x treatment interactions in practice, at least in the U.S.A. Empirical and conventional Bayes methods provide attractive alternatives to conventional ANOVAs for analysing and reporting the findings from multi-centre trials and do not require more restrictive assumptions than the ANOVA approach. These approaches require regarding the centre effects as random instead of fixed, a view which often will reasonably describe outcomes of clinical trials in spite of the fact that the individual centres certainly do not comprise a random sample of all possible centres. The components of these approaches are well understood and have been employed in related applications such as meta-analysis. Combining them in a way that makes their application to routine multi-centre trial analysis relatively straightforward does not appear to have been described previously, and is what forms the topic of this paper. The empirical Bayes approach leads to useful graphical displays, including one with the data superimposed on probability contours of the joint distribution of the individual centre means and standard deviations, which provides a handy way to identify possible outliers. Covariates can be incorporated without difficulty. The Bayes approach, implemented with Gibbs sampling, provides a convenient way to construct posterior and predictive distributions for a variety of useful statistics. We compare the result of empirical and conventional Bayes analyses with the result of fixed and mixed model ANOVAs applied to data from a multi-centre trial.

Cholesterol reduction yields clinical benefit: impact of statin trials

BACKGROUND

We determined the effect of incorporating the results of eight recently published trials of Hmg CoA reductase inhibitors ("statins") on the conclusions from our previously published meta-analysis regarding the clinical benefit of cholesterol lowering.

METHODS AND RESULTS

We used the same analytic approach as in our previous investigation, separating the specific effects of cholesterol lowering from the effects attributable to the different types of intervention studied. The reductions in coronary heart disease (CHD) and total mortality risk observed for the statins fell near the predictions from our earlier meta-analysis. Including the statin trial findings into the calculations led to a prediction that for every 10 percentage points of cholesterol lowering, CHD mortality risk would be reduced by 15% (P<.001), and total mortality risk would be reduced by 11% (P<.001), as opposed to the values of 13% and 10%, respectively, reported previously. Cholesterol lowering in general and by the statins in particular does not increase non-CHD mortality risk.

CONCLUSIONS

Adding the results from the statin trials confirmed our original conclusion that lowering cholesterol is clinically beneficial. The relationships (slope) between cholesterol lowering and reduction in CHD and total mortality risk became stronger, and the standard error of the estimated slopes decreased by about half. Use of statins does not increase non-CHD mortality risk. The effect of the statins on CHD and total mortality risk can be explained by their lipid-lowering ability and appears to be directly proportional to the degree to which they lower lipids.

Modifying the design of ongoing trials without unblinding

Conventional group sequential designs provide an objective basis for reducing the sample size of a trial if the difference between the treatments is much more or much less than anticipated. The flexibility of group sequential designs can be enhanced by allowing the sample size to increase when the variability turns out greater than expected. This can be accomplished by examining the variability before unblinding the data at the first stage of the trial. Depending on the result of this examination, the trial may continue as planned, or the design may change in various ways, for example, by increasing the sample size or by changing the number of stages or the scheduling of the interim analyses. The effect of this adaptive flexibility on the error rates turns out as one would expect from the findings for fixed-sample designs: no material impact on the type I error rate and an effect on the power that depends on the final total sample size.

Discussion of individual bioequivalence by M.-L. Chen

We review and discuss some technical statistical issues and practical implementation issues associated with the use of individual as opposed to population average bioequivalence to express the relative bioavailabilities of alternative formulations of a drug. A number of promising methods for addressing individual bioequivalence have been described. Individual bioequivalence calculations can be done using standard crossover designs, although more sophisticated assessments that compare test-reference variability to reference-reference variability require more complex designs. However, more experience about the clinical implications of various degrees of individual bioinequivalence as expressed by various metrics should be accumulated before definitive regulations are set forth mandating the use of individual bioequivalence for expressing relative bioavailabilities.

Prostate volume predicts outcome of treatment of benign prostatic hyperplasia with finasteride: meta-analysis of randomized clinical trials

OBJECTIVES

Six randomized clinical trials have compared at least 1 year of 5 mg finasteride to placebo in the treatment of clinical benign prostatic hyperplasia (BPH). The findings for the 2601 men in these trials provide an opportunity to investigate the heterogeneity of the effects seen in the individual studies and to identify pretreatment predictors of outcomes as expressed by symptoms or peak urinary flow rates.

METHODS

A formal meta-analysis using an Empirical Bayes approach employed data from all finasteride studies which included the Phase III trials in North America and Internationally, the Prospect, Early Intervention, and SCARP trials, and the Veterans Administration Cooperative Study which compared terazosin, finasteride, and the combination of these two drugs. A pooled analysis was also undertaken on the combined dataset.

RESULTS

The effect of finasteride treatment on improvements in total symptom severity, frequency score, and peak urinary flow rate was consistent across all six trials and similar among men with similar prostate volumes at baseline. Symptom severity improved by 1.8 points (95% confidence interval [CI], 0.7 to 2.9) in men with prostate volumes less than 20 cc (n = 72), while the improvement was 2.8 points (95% CI, 2.1 to 3.5) for men with volumes greater than 60 cc (n = 272) on the Quasi-IPSS Scale (range 0 to 30). Similarly, improvements in peak urinary flow rate ranged from 0.89 mL/s (95% CI, -0.05 to 1.83) for men with prostate volumes less than 20 cc to 1.84 mL/s (95% CI, 1.37 to 2.30) in men with volumes greater than 60 cc. The difference in the magnitude of improvement between finasteride and placebo becomes significant (that is, no overlap in 95% CI) for men with a baseline prostate volume assessed by either transrectal ultrasonography or magnetic resonance imaging of greater than 40 cc, which encompasses approximately 50% of the entire population. Baseline prostate volume is a key predictor of treatment outcomes: approximately 80% of the variation in the treatment effects noted between studies could be attributed to differences in mean prostate volumes at baseline. Variation in entry criteria results in large differences in baseline symptom severity status, prostate volume, and consequently apparent inconsistencies in the overall outcomes of these trials.

CONCLUSIONS

This meta-analysis suggests that finasteride is most effective in men with large prostates. Men with small prostates may not be suitable candidates for finasteride therapy for BPH. The need for a careful reevaluation of the definitions and terminology used when discussing urination problems is apparent.

Re-evaluating design specifications of longitudinal clinical trials without unblinding when the key response is rate of change

The design of clinical trials often requires knowledge of quantities such as between- and within subject variances about which only imprecise information exists. To provide assurance that the study has the desired power to detect a minimum clinically meaningful difference between treatment groups. Gould, Gould and Shih, and Shih have recommended obtaining relevant information from the trial at an interim stage without unblinding. Wittes and Brittain provided a similar recommendation, but viewed the portion up to the interim stage as an (internal) pilot study and required unblinding. This paper considers the problem of re-evaluating the design specifications in longitudinal clinical trials when the key response is the rate of change (slope). The proposed method aims to re-evaluate the sample size and study duration in a way that maintains the trial's blinding, using an EM algorithm. Simulation results show that the effect on type I error rate in negligible, but the potential gain in power can be substantial. The procedure is simple to use in practice, as it does not unblind patients' treatment identifications, and, since it does not unveil the relative efficacy of treatments, it fulfils the requirement of a valid 'administrative' (interim) analysis.

Planning and revising the sample size for a trial

The sample size for a trial depends on the type I and type II error rates and on the minimum relevant clinical difference, all of which are known, and on the anticipated, but unknown, value of a measure of variation for the key response. This measure is the overall response rate when the key response is binomially distributed, or the residual variance in each treatment group when the key response is continuous and normally distributed. Since the true value of the measure is unknown, it must be guessed or estimated from previous trials. We describe approaches to determine an appropriate value for it, both before the trial begins and after it has begun, for use in calculating the final sample size. These approaches differ from previously described 'internal pilot' methods in not requiring unblinding of the treatment assignments in the trial. They preserve the power and do not affect the type I error rate materially. The approaches can be applied to longitudinal studies where the rate of change over time is the response of interest, and to group sequential trials.

Cholesterol reduction yields clinical benefit. A new look at old data

BACKGROUND

There has been a continuing debate about the overall benefit of cholesterol lowering. We performed a novel meta-analysis of all randomized trials of more than 2 years' duration (n = 35 trials) to describe how coronary-heart-disease (CHD), non-CHD, and total mortality are related to cholesterol lowering and to type of intervention.

METHODS AND RESULTS

The analytic approach was designed to separate the effects of cholesterol lowering itself from the other effects of the different types of intervention used. For every 10 percentage points of cholesterol lowering, CHD mortality was reduced by 13% (P < .002) and total mortality by 10% (P < .03). Cholesterol lowering had no effect on non-CHD mortality. Certain types of intervention had specific effects independent of cholesterol lowering. Fibrates (clofibrates, 7 trials; gemfibrozil, 2 trials) increased non-CHD mortality by about 30% (P < .01) and total mortality by about 17% (P < .02). Hormones (estrogen, 2 trials; dextrothyroxin, 2 trials) increased CHD mortality in men by about 27% (P < .04), non-CHD mortality by about 55% (P < .03), and total mortality by about 33% (P < .01). No specific effects independent of cholesterol lowering were found due to diet (n = 11) or other interventions (resins, 5; niacin, 3; statins, 2; partial ileal bypass, 1).

CONCLUSIONS

The results suggest that cholesterol lowering itself is beneficial but that specific adverse effects of fibrates and hormones increase the risk of CHD (hormones only), non-CHD, and total mortality.

Group sequential extensions of a standard bioequivalence testing procedure

Bioequivalence trials compare the relative bioavailability of different formulations of a drug. Regulatory requirements for demonstrating average bioequivalence of two formulations generally include showing that a (say) 90% confidence interval for the ratio of expected pharmacologic end point values of the formulations lies between specified end points, e.g., 0.8-1.25. The likelihood of demonstrating bioequivalence when the formulations truly are equivalent depends on the sample size and on the variability of the pharmacologic end point. Group sequential bioequivalence testing provides a statistically valid way to accommodate misspecification of the variability in designing the trial by allowing for additional observations if a clear decision to accept or reject bioequivalence cannot be reached with the initial set of observations. This paper describes group sequential bioequivalence designs applicable in most practical situations that allow a decision to be reached with fewer observations than fixed-sample designs about 60% of the time at approximately the same average cost. The designs can be used in trials where the formulations are expected to have equal bioavailability and in trials where the formulations are expected to differ slightly. Data analyses are carried out exactly as for fixed-sample designs. Providing the capability of sequential decisions modestly affects the nominal significance levels, e.g., the required confidence level may be 93-94% instead of 90%.

Sample sizes for event rate equivalence trials using prior information

Trials for demonstrating the 'equivalence' of active standard and test treatments generally require large sample sizes that depend on the definition of 'equivalence' and the overall event rate when the outcome is incidence of an event such as mortality. The planning of sample sizes for such trials requires specification of a value for the overall event rate. This value often will reflect the outcomes of previous trials of the standard treatment, and is subject to uncertainty that needs some accommodation, to protect against an inadequate sample. Bayes and Empirical Bayes methods can be used to incorporate information from one or more previous trials into the sample size calculation when equivalence means high confidence that the event rate ratio is less than some specified value.

Monitoring of clinical trials and interim analyses from a drug sponsor's point of view

This paper illustrates aspects of data monitoring of clinical trials in the pharmaceutical industry. Formal interim analyses are performed at least in part to address the question of whether the trial should proceed or whether there should be an early termination of the trial. For formal interim analyses, frequently independent data and safety monitoring committees are utilized for monitoring clinical trials, and adjustments to nominal significance levels for test statistics are required. Various statistical methods developed during the last fifteen years are utilized. Administrative interim analyses are those analyses that are performed without any intention to stop the trial as a consequence of those analyses. For administrative interim analyses, adjustments to significance levels may not be required, but results must still be carefully interpreted. Regardless of the interim analyses performed, it is critical that the plans for interim analyses be identified in the study protocol, and the dissemination of interim results be carefully restricted. The following clinical trials sponsored by Merck Sharp and Dohme Research Laboratories (MSDRL) will illustrate these points: CONSENSUS; CONSENSUS II; 4S; Haemophilus influenza type b efficacy trial; famotidine in upper gastrointestinal haemorrhage, and a phase II analgesic study. It is anticipated that data monitoring and interim analysis activities will increase for future clinical trials due to the availability of appropriate statistical methods and improved data management systems.

Interim analyses for monitoring clinical trials that do not materially affect the type I error rate

Monitoring clinical trials often requires examining the interim findings to see if the sample size originally specified in the protocol will provide the required power against the null hypothesis when the alternative hypothesis is true, and to increase the sample size if necessary. This paper presents a new method, based on the overall response rate, for carrying out interim power evaluations when the observations have binomial distributions, without unblinding the treatment assignments or materially affecting the type I error rate. Simulation study results confirm the performance of the method.

Another view of active-controlled trials

Placebo-controlled efficacy trials may become more difficult to carry out with the increasing availability of effective therapies, especially for serious illnesses where denial of effective therapy may be objectionable ethically. Active-controlled trials aimed at establishing efficacy by demonstration of "equivalence" to "standard" therapy have potentially serious interpretational problems, and do not necessarily encourage good experimental practice. This article describes an alternative approach to the analysis of data from active-controlled trials using the information that makes an active-controlled trial necessary or desirable, namely a large, valid body of information about the consequence of using placebo. The approach uses information about placebo responses and also active agent responses from prior placebo-controlled trials to determine the likelihood of a significant active-placebo difference in an active-controlled trial, or in a trial with a vestigial placebo group. The sensitivity of the treatment comparisons depends directly on the quality of the design and execution of the active-controlled trial. The method is illustrated with data from trials of an H2-receptor antagonist in the treatment of acute duodenal ulcer.

Expanded clinical evaluation of lovastatin (EXCEL) study results. II. Assessment of the human lens after 48 weeks of treatment with lovastatin

The crystalline lenses of hypercholesterolemic patients were assessed before and after 48 weeks of treatment with lovastatin or placebo to determine the effect of lovastatin on the human lens. Patients were given a biomicroscopic (slit-lamp) examination of the lens, and a previously validated, standardized classification system was used to describe the findings. A total of 8,245 patients were randomly assigned in equal numbers to treatment with placebo or lovastatin 20 or 40 mg once or twice daily in this double-blind, parallel-group study. Statistical analyses of the distribution of cortical, nuclear and subcapsular opacities at 48 weeks, adjusted for age and presence of an opacity at baseline, showed no significant differences (p less than 0.01) between the placebo and lovastatin-treated groups. Visual acuity assessments at week 48 were also not found to have significantly different distributions among treatment groups. Moreover, no significant differences were found among the groups in the frequencies of greater than or equal to 2-line worsening in visual acuity with concurrent progression in lenticular opacity, cataract extraction, or any spontaneously reported adverse ophthalmologic experience. No evidence was found for an effect of lovastatin on the human lens after 48 weeks of treatment.

Expanded Clinical Evaluation of Lovastatin (EXCEL) study results. I. Efficacy in modifying plasma lipoproteins and adverse event profile in 8245 patients with moderate hypercholesterolemia

In the Expanded Clinical Evaluation of Lovastatin (EXCEL) Study, a multicenter, double-blind, diet- and placebo-controlled trial, we evaluated the efficacy and safety of lovastatin in 8245 patients with moderate hypercholesterolemia. Patients were randomly assigned to receive placebo or lovastatin at a dosage of 20 mg once daily, 40 mg once daily, 20 mg twice daily, or 40 mg twice daily for 48 weeks. Lovastatin produced sustained, dose-related (P less than .001) changes as follows (for dosages of 20 to 80 mg/d): decreased low-density lipoprotein-cholesterol level (24% to 40%), increased high-density lipoprotein-cholesterol level (6.6% to 9.5%), decreased total cholesterol level (17% to 29%), and decreased triglyceride level (10% to 19%). The National Cholesterol Education Program's low-density lipoprotein-cholesterol level goal of less than 4.14 mmol/L (160 mg/dL) was achieved by 80% to 96% of patients, while the less than 3.36 mmol/L (130 mg/dL) goal was achieved by 38% to 83% of patients. The difference between lovastatin and placebo in the incidence of clinical adverse experiences requiring discontinuation was small, ranging from 1.2% at 20 mg twice daily to 1.9% at 80 mg/d. Successive transaminase level elevations greater than three times the upper limit of normal were observed in 0.1% of patients receiving placebo and 20 mg/d of lovastatin, increasing to 0.9% in those receiving 40 mg/d and 1.5% in those receiving 80 mg/d of lovastatin (P less than .001 for trend). Myopathy, defined as muscle symptoms with a creatine kinase elevation greater than 10 times the upper limit of normal, was found in only one patient (0.1%) receiving 40 mg once daily and four patients (0.2%) receiving 80 mg/d of lovastatin. Thus, lovastatin, when added after an adequate trial of a prudent diet, is a highly effective and generally well-tolerated treatment for patients with moderate hypercholesterolemia.

Applying survival methodology to adverse experience occurrences in controlled clinical trials

On the basis of calculated cumulative hazard rates for initial occurrence of adverse experiences of patients following treatment for rheumatoid arthritis and osteoarthritis, a simple function is evolved that fits such cumulative hazard rate data very well. From this simple function, we obtain the estimated hazard rate in terms of two physically meaningful parameters. These two parameters can be used to describe the rate of occurrence of adverse experiences, and to convey the concept of risk of adverse experience associated with duration of exposure to a drug. The parameters are rho 1, which represents the risk of adverse experience at baseline, and rho 2, which defines the rate of occurrence of adverse experience immediately following drug administration. A method of estimating rho 1 and rho 2 is the maximum likelihood approach, and the estimated parameters are given for series of data referring to patients treated for arthritis.

The analysis of titration studies in phase III clinical trials

Clinical trials commonly employ the titration design for certain drugs such as antihypertensives. In a Phase III trial the design has purposes distinct from those of a Phase I or II trial, as well as from those of a trial with a parallel design. In this paper we compare the titration design with the usual parallel design in their respective purposes for Phase III trials, explore the relevant questions addressed, and examine typical data from such trials. We also discuss work which focuses primarily on the Phase I or II titration trials. We formulate the problem in the framework of one-way contingency table augmented with incomplete data and obtain the maximum likelihood estimates of the parameters and their estimated variances/covariances via the EM algorithm. An example of a Phase III study of an antihypertensive agent illustrates the proposed procedure.

Effect of cancer chemotherapy on the immune response to influenza virus vaccine: review of published studies

Controversy exists regarding the ability of cancer chemotherapy to prevent the development of an adequate immune response to influenza virus vaccine. Of 12 studies addressing this issue, eight demonstrated a significant lessening of the immune response among patients receiving cancer chemotherapy. The other four studies failed to find a significant difference between the immune responses of patients receiving cancer chemotherapy and persons not receiving chemotherapy; for these studies the Type 2 error rate (i.e., the probability of wrongly concluding that no difference exists) was calculated. Since the response rates in the four inconclusive studies were consistent with those in the other eight studies but the sample sizes were much smaller, the failure of the former studies to find significant differences in immune responses probably was due to insensitivity rather than to the absence of such differences. The preponderance of evidence suggests that the serum antibody response to influenza virus vaccine is significantly weaker in patients receiving cancer chemotherapy than in persons not receiving chemotherapy.

A new approach to the analysis of clinical drug trials with withdrawals

In clinical drug trials which require observation of patients for more than a few days or weeks, some patients may withdraw before completing the planned course of the trial for reasons that are related to the therapy, for example adverse experiences or lack of therapeutic effect. If the number of such withdrawals is material, then analysis of the data from these trials should account for these withdrawals. Ignoring them in the analyses or using the last recorded value before withdrawl can lead to misleading conclusions. If the response outcomes can be ordered so that, for instance, adverse withdrawals can be regarded as "bad" outcomes, then the withdrawals can be incorporated into the analyses straightforwardly. The method is described and illustrated, and several issues arising in its application are discussed.

Dexamethasone and severe head injury. A prospective double-blind study

A prospective double-blind study of the effects of dexamethasone administration on the outcome of patients with severe head injuries was performed. Patients were stratified for severity of neurological injury and were treated with placebo, low-dose dexamethasone (16 mg/day), or high-dose dexamethasone (96 mg/day) for a period of 6 days. Outcome was evaluated at 6 months following injury. Of the 76 patients available for analysis, a good outcome was achieved in 37% of placebo-treated patients, 44% of low-dose-treated patients, and 29% of high-dose-treated patients. These differences are not statistically significant. Similarly dexamethasone administration had no statistically significant effect on intracranial pressure patterns or serial neurological examinations during hospitalization. Gastrointestinal bleeding occurred in only one patient. Good outcome was associated with age under 10 years, lighter depth of coma on admission, and the preservation of brain-stem reflexes upon admission. A recalculation of data in previous clinical series purporting to show an improvement in outcome as a result of corticosteroid therapy shows no significant difference in outcome when steroid- and placebo-treated patients are compared. In our series, 90% of all deaths were caused by recurrent intracranial hematomas, medical complications, or diffuse brain injuries with parenchymal hemorrhage and tissue disruption -- causes of death which cannot be affected by corticosteroid therapy. The study suggests that dexamethasone in either high or low dosages has no significant effect on morbidity and mortality following severe head injury.

Safety of cefoxitin: an approach to the analysis of laboratory data

The safety of cefoxitin, in terms of values obtained in laboratory tests during and after therapy, was estimated by three methods for analysis of data derived from controlled clinical comparisons of cephalothin and cefoxitin. Both antibiotics were found to be safe with respect to hematologic, renal, and hepatic function and did not differ significantly from each other. Laboratory data confirmed by tests performed serially and by paired related tests were analyzed by a novel method of comparison.

Intraocular pressure decrease in normal volunteers following timolol ophthalmic solution

Timolol ophthalmic solutions 0.5 per cent, 1.0 per cent, and 1.5 per cent lowered intraocular pressures significantly in normal human volunteers. Maximum lowering of the intraocular pressures was reached at two hours with the 0.5 per cent solution of timolol and at one hour with the 1.0 per cent and 1.5 per cent timolol ophthalmic solutions. The effect lasted the full seven hours of observations. No objective or subjective evidence of ocular irritation could be attributed to the drug. A single dose of timolol applied topically to the eyes of normal human volunteers had no effect on pupillary size, visual acuity, blood pressure, or pulse rate.