PLoS One. 2013; 8(3): e59104.

Statistical AnalysisCorrelations between RKIP expression and clinical data from patients were performed using the chi-square test (χ2-test). Cumulative survival probabilities were calculated using the Kaplan-Meier method. Differences between survival rates were tested with the log-rank test. The statistical analysis was performed using SPSS software for Windows, version 17.0. For in vitro and in vivo assays, single comparisons between the different conditions studied were done using Student’s t test, and differences between groups were tested using two-way analysis of variance (ANOVA). Statistical analysis was done using Graph Pad Prism version 5. The level of significance in all the statistical analysis was set at p<0.05.

BMC Fam Pract. 2013; 14: 45.

Sensitivity analysesSensitivity analyses were conducted to test the robustness of the results. In the first sensitivity analysis the costs for the second year were discounted with 4% and QALYs achieved in this year were discounted with 1.5%, in line with Dutch guidelines [24]. In the second sensitivity analysis the productivity losses were valued with the human capital method. In the third sensitivity analysis the costs for sports were excluded. In the fourth sensitivity analysis, missing data for follow-up phone calls (which in the base case analysis were assumed to be zero) were multiply imputed. For the fifth sensitivity analysis, 50 multiply imputed files were created, according to the rule of thumb that the number of imputed files should at least equal the fraction of incomplete cases [25]. The sixth sensitivity analysis was restricted to participants with complete cost and effect data, i.e., complete case analysis (CCA). In the final sensitivity analysis, cost-effectiveness was assessed from the healthcare perspective.

Ann Surg. 2007 June; 245(6): 831–842.

At the time of planning, it was decided that a sample size of 40 patients per treatment group would provide the study with a statistical power of 80% at the 0.05 level of significance to detect a reduction in recurrence rate from 60% to 30%.20 The sample size per treatment group remained unchanged when the protocol was revised in the year 2000.

PLoS One. 2013; 8(1): e53719

. The assumptions we made for calculation of the sample size were based on data from several uncontrolled studies reporting significant improvement in liver function tests following bone marrow cell therapy [20] [31], and on the histological evidence of a sharp increase in hepatic progenitors proliferative activity early after G-CSF administration [15]. Patients who missed the 3-month follow-up visit, died or were transplanted were considered as not having reached the primary endpoint. Due to the non-parametric distribution of variables, data were expressed as mean and its standard deviation or median and its range. The Fisher’s exact test was used to compare the proportion of patients in each group reaching the primary endpoint. Binomial methods were used to construct the confidence intervals around percentages. Quantitative variables between groups were compared using the non-parametric Mann-Whitney U-test. The ANOVA repeated measure test with Bonferroni correction for multiple comparisons was used to assess MELD changes in each treatment group for the effect of time. Cytokines values and liver tissue parameters at baseline and at 3 months were compared using the Wilcoxon signed rank test. Correlations between MELD score changes and bone marrow cell subpopulations were made using the Spearman rank test. All analyses were based on the intention-to-treat principles. All tests were two-sided, and the level of significance was set at a p<0.05. The analyses were conducted using Statistical Package for the Social Sciences (SPSS 10.0 Chicago, IL, USA).

PLoS One. 2013; 8(1): e53719

Sample Size and Statistical AnalysisTo achieve a statistical power of 85% at 5% type I error, with the initial assumption that the primary endpoint will be reached in 60% of BMMCT-treated patients and 20% in patients on SMT alone, a sample size of 26 patients in each arm was calculated. Assuming a 10% of included patients lost to follow-up, a sample of 60 patients was required.

Am J Gastroenterol. 2013 August; 108(8): 1305–1313.

Statistical analysisThis study was designed to detect a 15% increase in 2-year recurrence-free survival in the CTHA/CTAP group from an anticipated 35% in the control group. To detect this difference with a power of 80% and type I error of 5% (two-sided test), we needed 280 patients (140 for each arm). Differences between groups for each characteristic were tested for significance with Fisher's exact test for categorical variables and t-test for continuous variables. All data necessary for analysis was corrected in the main computer server system of University of Tokyo, Department of Gastroenterology.Recurrence-free survival and overall survival were calculated using the Kaplan–Meier method and were compared by the log-rank test. Cox proportional hazard regression was used to calculate hazard ratios with 95% confidence interval (CI) between the groups in univariate and multivariate settings. The primary end point was evaluated in subgroups according to the following characteristics: age, sex, body mass index, treatment naivety, hepatitis B surface antigen (HBsAg) positivity, hepatitis C virus antibody positivity, tumor size, tumor number, platelet count, tumor marker positivity for α-fetoprotein (AFP), lens culinaris agglutinin-reactive fraction of AFP, and des-γ-carboxy prothrombin. An adjusted hazard ratio comparing the groups was calculated using multivariate Cox regression with factors that showed significance in univariate analysis. Data at entry were used for the analyses. A post hoc analysis comparing the recurrence-free survival of those with and without newly diagnosed HCC in the CTHA/CTAP group was performed.All analyses were performed on an intention-to-treat basis. Differences with a two-sided P value of <0.05 were considered statistically significant. Data processing and analysis were performed with S-PLUS ver. 7 (TIBCO Software, Palo Alto, CA). Finally, all authors had access to the study data and had reviewed and approved the final manuscript.

PLoS One. 2013; 8(1): e54676.

Production of Simvastatin Drug Delivery SystemSimvastatin solution (Watanabe Chemical Co., Osaka, Japan) at a concentration of 4 mg/mL were immersed with the β-TCP samples in a rotaevaporator (Buchi Rotavapor RT200) until the solution were dried and subsequently placed in a 100% humidity vacuum seal. The simvastatin loaded β-TCP were further coated with an apatite outer layer. The apatite cement bulk powder consist of equimolar mixture of tetratricalcium phosphate (TTCP) and dicalcium phosphate dihydrate (DCPD) (Wako, Tokyo) and was prepared by grinding at 20 per second for 17 mins in an agate vibration mixer mill (Retsch Co., Germany; 10 mL volume chamber in a ball 10 mm in diameter). This cement bulk powder (0.470 g) was poured into a silicon rubber mould (5 mm in diameter with 2 mm thickness) for 1 hour, and stored at room temperature in a vacuum seal with 100% relative humidity for 24 hrs.

PLoS One. 2013; 8(1): e54676.

Evaluation of in-vitro Degradation of β-TCPThe in-vitro degradation of the β-TCP looking specifically at the release of calcium and magnesium was evaluated in simulated body fluid solution (SBF) [15]. The samples were each immersed in 5 mL of the buffer solution and placed in a shaking water bath at 37°C. At each predetermined time point, the buffer solution were collected and replaced with fresh buffer every 24 hours for 7 days. The collected solutions were than evaluated by ICP-MS.

PLoS One. 2013; 8(1): e54676.

Physico-chemical CharacterizationThe powder X-ray diffraction (XRD) profiles of the coral before and after hydrothermal conversion were measured by powder XRD analysis (RINT- Ultima-III, Rigaku Co., Japan; CuKα radiation, 40 kV, 40 mA). The step scanning was performed with an integration time of 1 min at intervals of 2° (2θ) and matched with JCPDS database. The chemical composition of the crushed sample powder was investigated by fourier transform infrared spectroscopy (FTIR). Samples were ground with 1% KBr in an agate mortar, and analyzed under nitrogen atmosphere from 2000 to 400 cm−1 using a Nicolet IR 760. Inductively coupled plasma-mass spectroscopy (ICP-MS) was used to measure the chemical composition of the samples by using approximately 0.3 g of sample which was digested with 0.25 mL of HNO3 and 0.25 mL of H2O2. Once the digestion was completed the sample volume was made up to 5 mL with H2O. The samples underwent a further 1100 dilution before ICP-MS analysis. Samples were diluted further as needed. The surface morphology was characterized by scanning electron microscopy (JEOL JSM-7600F, Field Emission SEM, 10 KV). The internal architectural structure was characterized by a micro-CT scanner (InspeXio; Shimadzu Science East Corporation, Tokyo, Japan) with a voxel size of 70 mm/pixel as a non-destructive method. Tri/3D-Bon software (RATOC System Engineering Co. Ltd, Tokyo, Japan) was used to generate a complete 3D reconstruction of the sample. The surface area was measured by using a Quantachrome Monosorb™ B.E.T. surface area analyzer and the pore size distribution profile was measured by nitrogen volumetric adsorption measurements (Quantachrome Autosorb pore size analyser).

PLoS One. 2013; 8(1): e54676.

Hydrothermal Synthesis of β-TCP StarsForaminifera samples were purchased commercially from Business Support Okinawa Co. Ltd., Japan. The samples were first cleansed in sodium hydrochlorite for 20 mins and dried at 40°C for 2 hours and placed in a heating oven at 220°C for 48 hours with aqueous diammonium hydrogen phosphate [(NH4)2(HPO4)] (Wako Chemical Co., Tokyo, Japan). The diammonium hydrogen phosphate solution was adjusted to yield Ca/P molar ratios of 1.5 to produce β-TCP. The resulting samples were than subsequently characterized by the following methods. Furthermore no specific permits were required for the described field studies.

PLoS One. 2012; 7(2): e31807.

Statistical analysisData are presented as the mean±standard deviation (SD). Statistical analysis was performed using one-way analysis of variance (ANOVA) followed by SNK tests as a post hoc test. The Kruskal-Wallis test was used to evaluate the differences in categorical values followed by Mann-Whitney U tests as a post hoc test. p value of <0.05 was accepted as statistically significant.

PLoS One. 2013; 8(6): e67242.

Statistical AnalysisAll quantitative data are presented as the mean ± SEM. According to normality test results, data were compared using a one-way ANOVA followed by a LSD post hoc test. Two-tailed P-values < 0.05 were considered significant and adjusted P-values were used among the subgroup comparison analyses. All statistical analyses were performed with the software package SPSS 16.0 (IBM, USA) for Windows.

Diabetol Metab Syndr. 2012; 4: 36.

For the short-term period, a primary efficacy analysis was performed for the endpoint of change in HbA1c from baseline to week 24 comparing the saxagliptin q.A.M. treatment groups to placebo, with missing data imputed on a last-observation-carried-forward (LOCF) basis. A two-step, parallel gatekeeping methodology was used to preserve the overall type I error rate at the 0.05 level [8]. As a first step, comparisons were made between the saxagliptin 2.5mg q.A.M. and 5mg q.A.M. treatment groups versus the placebo group. If either saxagliptin group showed significance for HbA1c at the α=0.027 level versus placebo, the second step comparison would use the same α level; if both first-step comparisons reached significance, α=0.05 would be used. The second comparison was the saxagliptin 2.5/5mg q.A.M. titrated group versus placebo. Assuming a standard deviation (SD) of 1.1%, 62 patients per treatment group would provide 90% power to detect a difference in means of 0.7% between saxagliptin 2.5mg q.A.M. or 5mg q.A.M versus placebo.

Lancet Oncol. 2013 July; 14(8): 749–759.

Statistical analysisUnder the original design, we aimed to detect a 25% reduction in hazard rate (80% power; 5% significance level; two-sided log-rank test) for the primary endpoint, overall survival, with the addition of panitumumab to irinotecan. Anticipated median overall survival with irinotecan was 9 months,11 with a targeted improvement to 12 months with the addition of panitumumab, resulting in a sample size of 720 patients and at least 380 deaths.In the amended design, we anticipated an increased treatment benefit with IrPan in the refined primary population of KRASc.12,13,61 wild-type patients not pre-treated with EGFR monoclonal antibodies. We have previously assessed KRAS as a prognostic and predictive marker in patients treated with cytotoxic chemotherapy alone,10 and on the basis of these data, we made no change to the predicted overall survival of 9 months for KRAS wild-type patients in the irinotecan alone group. However, in the new design we aimed to detect a 30% reduction in hazard rate, corresponding to a median overall survival of 12·9 months with the addition of panitumumab. Target accrual was 466 patients in the primary population, with the analysis planned after at least 246 deaths had occurred. An interim analysis was planned to address inferiority or superiority of irinotecan plus panitumumab compared with irinotecan alone, with a stringent p value of 0·001, therefore no adjustment was required in the final significance level.12 Secondary endpoints included progression-free survival (PFS), the proportion of patients who achieved a RECIST response, quality of life, and toxicity. Post-hoc statistical comparisons were made between the rates of grade 3 or higher events in the two groups, using univariate χ2 tests (or Fisher's exact test for five or fewer events) at the 5% significance level. This analysis did not account for multiple testing and its findings should be interpreted with caution.We had two predefined exploratory populations: patients with KRASc.12,13,61-mutated tumours randomised to irinotecan versus IrPan before the protocol modification; and patients previously treated with an anti-EGFR monoclonal antibody.Additional analyses were later planned, before final analysis, to investigate any interaction between BRAFc.600, NRASc.12,13,61, KRASc.146, or PIK3CA status and the effect of panitumumab. In planning these analyses, molecular subgroups were predefined to determine treatment interaction with mutation status, with the pre-existing hypothesis that KRASc.12,13,61 wild-type patients with a mutation at one of the other loci would have less benefit from panitumumab than would patients with no mutations. Patients were grouped as having any mutation (a mutation at any other one of the assessed loci) or as all wild-type (no mutations at the loci tested). In the analysis, missing data for an individual gene was imputed as wild-type, but we did a sensitivity analysis in which only patients confirmed to be wild-type at all 12 loci were classed as all wild-type. We did a second sensitivity analysis excluding PIK3CA mutation from the analysis.For individual rare mutations occurring in less than 10% of patients, PICCOLO provides only minimal power (about 10%) to detect clinically significant treatment effects (eg, reduction in hazard rate of 30%). These analyses are therefore exploratory in nature and should not be over-interpreted. Cox's proportional hazards modelling, adjusting for minimisation factors, was pre-specified for overall survival and PFS. Statistical testing was post hoc for response rate and toxicity.Primary analysis of all endpoints was scheduled after 246 deaths, as per the amended trial design. On recommendation from the data monitoring and ethics committee, we also planned a final updated analysis of overall survival when at least 2 years had passed since all patients were allocated to treatment. We report here the primary event-driven overall survival analysis in the primary population. We also report the secondary endpoints and final analysis of overall survival, in the primary population, its planned molecular subgroups, and in the exploratory population of patients with mutations at KRASc.12,13,61. Results in patients previously treated with an anti-EGFR monoclonal antibody will be reported elsewhere, as will results for the comparison of irinotecan versus irinotecan plus ciclosporin. We used SAS (version 9.2) for all statistical analyses.This study is registered as an International Standard Randomised Controlled Trial, number ISRCTN93248876.

Lancet Oncol. 2010 May; 11(5): 421–428.

Statistical analysisWe estimated that 60–80% of the patients at our centre with locally advanced breast cancer would have DTCs at baseline, and that neoadjuvant chemotherapy would decrease this by about 20% at 3 months (similar to the percentage of pCR expected in this population). The additional benefit attributable to zoledronic acid was unknown; however, we estimated this effect to be roughly equivalent to doubling the pCR rate. The trial was designed to have at least 80% power at a 0·05 significance level to detect a 20–26% difference in DTCs at baseline versus 3 months, in patients with and without zoledronic acid therapy, even if no DTCs were detected at baseline. Treatment with zoledronic acid was expected to show increasing bone-mineral density at 12 months and decreasing concentrations of bone-turnover markers at 3 months and 12 months compared with the non-zoledronic acid group, as in previous studies of women with low bone-mineral density.9Patients’ baseline characteristics were analysed with descriptive statistics. Categorical data, including DTC analysis, were described using frequencies and percentages and tested using χ² and one-sided Fisher’s exact tests. DTC analysis included all patients with data at each timepoint. All patients were treated according to the group into which they were randomly assigned.Continuous data were described using means and standard deviations for normally distributed variables, and otherwise by median and percentiles. Patients with analysable bone-turnover markers were defined as those with a baseline and either a 3-month or 12-month measurement. Patients were defined as analysable for bone-mineral density if they had two determinations. Paired t tests were done for change from baseline in bone-turnover markers and bone-mineral density. Comparison across treatment groups was done using an unpaired t-test. Multiple linear regression analysis was done to control for smoking, body-mass index (BMI), and endocrine therapy in the analysis of bone-mineral density. Mixed models (random intercept and slopes) were used, with each of the bone-turnover markers as dependent variables, to model the treatment effect over time while adjusting for possible confounders. For analysis of bone-turnover markers, time was modelled as linear and the independent variables were BMI, use of adjuvant endocrine therapy, and smoking habits. All tests for bone-turnover markers and bone-mineral density were two-sided and an alpha of less than 0·05 was considered significant. SAS version 9.1 was used for statistical analysis. This study is registered with ClinicalTrials.gov, NCT00242203.