%TTD and %TUDD: New SAS macro programs to calculate the survival data of the time to deterioration for patient-reported outcomes data in oncology.

BACKGROUND AND OBJECTIVE: Longitudinal analysis of patient-reported outcome (PRO) data remains challenging, as no standardization of statistical methods has been proposed, making comparison of PRO results between clinical trials difficult. In this context, the time to deterioration approach has recently been proposed and is regularly used as a modality of longitudinal PRO analysis in oncology. METHODS: Two new SAS macro programs were developed, %TTD and %TUDD, which implement longitudinal analysis of PRO data according to the time to deterioration approach. These programs implement the recommended deterioration definitions. We described the programs with their different functionalities. RESULTS: The %TTD macro calculates the time to first or transient deterioration, and the %TUDD macro calculates the time until definitive deterioration. These macros allow to obtain the survival variables from the time to deterioration approach. We illustrate our programs by presenting different applications on the randomized phase II AFUGEM GERCOR clinical trial. CONCLUSION: The implementation of the deterioration definitions in SAS software allows the dissemination of this approach, in order to move toward the goal of standardization of longitudinal PRO analysis in oncology clinical trials.

Sequential first-line treatment with nab-paclitaxel/gemcitabine and FOLFIRINOX in metastatic pancreatic adenocarcinoma: GABRINOX phase Ib-II controlled clinical trial.

BACKGROUND: Nab-paclitaxel/gemcitabine (AG) and FOLFIRINOX (FFX) are promising drugs in metastatic pancreatic cancer (MPC). This study evaluated a new first-line sequential treatment (AG followed by FFX) in MPC that might overcome resistance to primary therapy and delay tumor progression. PATIENTS AND METHODS: Patients with histologically/cytologically confirmed MPC were included in a multicentric trial receiving AG (day 1, 8 and 15) followed by FFX (day 29 and 43). In phase Ib, three dose-levels were tested for maximum tolerated dose (MTD) and recommended phase II dose. In phase II, the main outcome was the objective response rate (ORR) and secondarily safety, progression-free survival (PFS) and overall survival (OS). RESULTS: In phase Ib, we included 33 patients (31 assessable) of median age 61.0 years (range 42-75 years) and represented by 54.8% males. Five dose-limiting toxicities were reported without any death. The main grade 3/4 toxicities were neutropenia with spontaneous resolution (35.5%/32.3%), venous thromboembolism (grade 3: 22.6%) and thrombopenia (grade 3: 29.0%), while the MTD was not reached. In phase II, we included 58 patients of median age 60 years (range 34-72 years), 50% males and with Eastern Cooperative Oncology Group stage score 0 and 1 of 37.9% and 62.1%, respectively. They received a median of 4 (1-9) cycles in 8.5 months (0.5-19.8 months). The ORR was 64.9% [95% confidence interval (CI) 51.1% to 77.1%], and neurotoxicity was remarkably low. The main grade 3-4 toxicities were venous thromboembolism, thrombopenia, neutropenia/febrile neutropenia, nausea, diarrhea, weight loss and asthenia without any death. Tumor response was complete in 3.5% and partial in 61.4%, while disease was stable in 19.3% and progressive in 15.8% of patients. The median PFS was 10.5 months (95% CI 6.0-12.5 months) and median OS was 15.1 months (95% CI 10.6-20.1 months). CONCLUSION: Sequential AG and FFX showed acceptable toxicity as first-line treatment with no limiting neurotoxicity, while high response rate and survival justify randomized trials.

Optimal biological dose: a systematic review in cancer phase I clinical trials.

BACKGROUND: Classical phase 1 dose-finding designs based on a single toxicity endpoint to assess the maximum tolerated dose were initially developed in the context of cytotoxic drugs. With the emergence of molecular targeted agents and immunotherapies, the concept of optimal biological dose (OBD) was subsequently introduced to account for efficacy in addition to toxicity. The objective was therefore to provide an overview of published phase 1 cancer clinical trials relying on the concept of OBD. METHODS: We performed a systematic review through a computerized search of the MEDLINE database to identify early phase cancer clinical trials that relied on OBD. Relevant publications were selected based on a two-step process by two independent readers. Relevant information (phase, type of therapeutic agents, objectives, endpoints and dose-finding design) were collected. RESULTS: We retrieved 37 articles. OBD was clearly mentioned as a trial objective (primary or secondary) for 22 articles and was traditionally defined as the smallest dose maximizing an efficacy criterion such as biological target: biological response, immune cells count for immunotherapies, or biological cell count for targeted therapies. Most trials considered a binary toxicity endpoint defined in terms of the proportion of patients who experienced a dose-limiting toxicity. Only two articles relied on an adaptive dose escalation design. CONCLUSIONS: In practice, OBD should be a primary objective for the assessment of the recommended phase 2 dose (RP2D) for a targeted therapy or immunotherapy phase I cancer trial. Dose escalation designs have to be adapted accordingly to account for both efficacy and toxicity.

Handling informative dropout in longitudinal analysis of health-related quality of life: application of three approaches to data from the esophageal cancer clinical trial PRODIGE 5/ACCORD 17.

BACKGROUND: Health-related quality of life (HRQoL) has become a major endpoint to assess the clinical benefit of new therapeutic strategies in oncology clinical trials. Typically, HRQoL outcomes are analyzed using linear mixed models (LMMs). However, longitudinal analysis of HRQoL in the presence of missing data remains complex and unstandardized. Our objective was to compare the modeling alternatives that account for informative dropout. METHODS: We investigated three alternative methods-the selection model (SM), pattern-mixture model (PMM), and shared-parameters model (SPM)-in relation to the LMM. We first compared them on the basis of methodological arguments highlighting their advantages and drawbacks. Then, we applied them to data from a randomized clinical trial that included 267 patients with advanced esophageal cancer for the analysis of four HRQoL dimensions evaluated using the European Organisation for Research and Treatment of Cancer (EORTC) QLQ-C30 questionnaire. RESULTS: We highlighted differences in terms of outputs, interpretation, and underlying modeling assumptions; this methodological comparison could guide the choice of method according to the context. In the application, none of the four models detected a significant difference between the two treatment arms. The estimated effect of time on HRQoL varied according to the method: for all analyzed dimensions, the PMM estimated an effect that contrasted with those estimated by the SM and SPM; the LMM estimated effects were confirmed by the SM (on two of four HRQoL dimensions) and SPM (on three of four HRQoL dimensions). CONCLUSIONS: The PMM, SM, or SPM should be used to confirm or invalidate the results of LMM analysis when informative dropout is suspected. Of these three alternative methods, the SPM appears to be the most interesting from both theoretical and practical viewpoints. TRIAL REGISTRATION: This study is registered with ClinicalTrials.gov , number NCT00861094 .

GUIP1: a R package for dose escalation strategies in phase I cancer clinical trials.

BACKGROUND: The main objective of phase I cancer clinical trials is to identify the maximum tolerated dose, usually defined as the highest dose associated with an acceptable level of severe toxicity during the first cycle of treatment. Several dose-escalation designs based on mathematical modeling of the dose-toxicity relationship have been developed. The main ones are: the continual reassessment method (CRM), the escalation with overdose control (EWOC) method and, for late-onset and cumulative toxicities, the time-to-event continual reassessment method (TITE-CRM) and the time-to-event escalation with overdose control (TITE-EWOC) methods. The objective of this work was to perform a user-friendly R package that combines the latter model-guided adaptive designs. RESULTS: GUIP1 is an R Graphical User Interface for dose escalation strategies in Phase 1 cancer clinical trials. It implements the CRM (based on Bayesian or maximum likelihood estimation), EWOC and TITE-CRM methods using the dfcrm and bcrm R packages, while the TITE-EWOC method has been specifically developed. The program is built using the TCL/TK programming language, which can be compiled via R software libraries (tcltk, tkrplot, tcltk2). GUIP1 offers the possibility of simulating and/or conducting and managing phase I clinical trials in real-time using file management options with automatic backup of study and/or simulation results. CONCLUSIONS: GUIP1 is implemented using the software R, which is widely used by statisticians in oncology. This package simplifies the use of the main model-based dose escalation methods and is designed to be fairly simple for beginners in R. Furthermore, it offers multiple possibilities such as a full traceability of the study. By including multiple innovative adaptive methods in a free and user-friendly program, we hope that GUIP1 will promote and facilitate their use in designing future phase I cancer clinical trials.

Time to deterioration in cancer randomized clinical trials for patient-reported outcomes data: a systematic review.

PURPOSE: The time to deterioration (TTD) approach has been proposed as a modality of longitudinal analysis of patient-reported outcomes (PROs) in cancer randomized clinical trials (RCTs). The objective of this study was to perform a systematic review of how the TTD approach has been used in phase III RCTs to analyze longitudinal PRO data. METHODS: A systematic literature search was conducted in PubMed/MEDLINE, the Cochrane Library and through manual search to identify studies published between January 2014 and June 2018. All phase III cancer RCTs including a PRO endpoint using the TTD approach were considered. We collected general information about the study, PRO assessment and the TTD approach, such as the event definition, the choice of reference score and whether the deterioration was definitive or not. RESULTS: A total of 1549 articles were screened, and 39 studies were finally identified as relevant according to predefined criteria. Among these 39 studies, 36 (92.3%) were in advanced and/or metastatic cancer. Several different deterioration definitions were used in RCTs, 10 studies (25.6%) defined the deterioration as "definitive", corresponding to a deterioration maintained over time until the last PRO assessment available for each patient. The baseline score was explicitly stated as the reference score to qualify the deterioration for most studies (n = 31, 79.5%). CONCLUSION: This review highlights the lack of standardization of the TTD approach for the analysis of PRO data in RCTs. Special attention should be paid to the definition of "deterioration", and this should be based on the specific cancer setting.

Longitudinal health-related quality of life analysis in oncology with time to event approaches, the STATA command qlqc30_TTD.

BACKGROUND AND OBJECTIVE: Health-related quality of life (HRQoL) has become one relevant and available alternative endpoint of clinical trials in cancer research to evaluate efficiency of care both for the patient and health system. HRQoL in oncology is mainly assessed using the 30-item European Organisation for Research and Treatment of Cancer Quality of Life-Questionnaire Core 30 (EORTC QLQ-C30). The EORTC QLQ-C30 questionnaire is usually assessed at different times along the clinical trials in order to analyze the kinetics of HRQoL evolution and to fully assess the impact of the treatment on the patient's HRQoL level. In this perspective, the realization of a longitudinal HRQoL analysis is essential and the time to HRQoL score deterioration approach is a method which is more and more used in clinical trials. METHOD: Using the Stata software, we developed a QLQ-C30 specific command, qlqc30_TTD, which implements longitudinal strategies based on the time to event methods by considering the time to HRQoL score deterioration. This user-written command providing automatic execution of the Time To Deterioration (TTD) and Time Until Definitive Deterioration (TUDD) methods. RESULT: The program implements all published definitions and provides the Kaplan-Meier curves for each dimension (by group) and a table with the Hazard Ratio and Log-Rank test. CONCLUSION: The longitudinal analysis of HRQoL data in cancer clinical trials remains complex with only few programs like ours computed. This program will be of great help and will allow a more systematic and quicker analysis of the HRQoL data in clinical trials in oncology.

Clinical utility of circulating DNA analysis for rapid detection of actionable mutations to select metastatic colorectal patients for anti-EGFR treatment.

BACKGROUND: While tumor-tissue remains the 'gold standard' for genetic analysis in cancer patients, it is challenged with the advent of circulating cell-free tumor DNA (ctDNA) analysis from blood samples. Here, we broaden our previous study on the clinical validation of plasma DNA in metastatic colorectal cancer patients, by evaluating its clinical utility under standard management care. PATIENTS AND METHODS: Concordance and data turnaround-time of ctDNA when compared with tumor-tissue analysis were studied in a real-time blinded prospective multicenter clinical study (n = 140 metastatic colorectal patients). Results are presented according to STARD criteria and were discussed in regard with clinical outcomes of patients. RESULTS: Much more mutations were found by ctDNA analysis: 59%, 11.8% and 14.4% of the patients were found KRAS, NRAS and BRAF mutant by ctDNA analysis instead of 44%, 8.8% and 7.2% by tumor-tissue analysis. Median tumor-tissue data turnaround-time was 16 days while 2 days for ctDNA analysis. Discordant samples analysis revealed that use of biopsy, long delay between tumor-tissue and blood collection and resection of the tumor at time of blood draw, tumor site, or type of tissue analyzed seem to affect concordance. Altogether, the clinical data with respect to the anti-epidermal growth factor receptor response (RAS status) and the prognosis (BRAF status) of those discordant patients do not appear contradictory to the mutational status as determined by plasma analysis. Lastly, we present the first distribution profile of the RAS and BRAF hotspot mutations as determined by ctDNA analysis (n = 119), revealing a high proportion of patients with multiple mutations (45% of the population and up to 5 mutations) and only 24% of WT scored patients for both genes. Mutation profile as determined from ctDNA analysis with using various detection thresholds highlights the importance of the test sensitivity. CONCLUSION: Our study showed that ctDNA could replace tumor-tissue analysis, and also clinical utility of ctDNA analysis by considerably reducing data turnaround time.

Health-related quality of life results from the PRODIGE 5/ACCORD 17 randomised trial of FOLFOX versus fluorouracil-cisplatin regimen in oesophageal cancer.

BACKGROUND: A recent prospective randomised trial did not reveal significant differences in median progression-free survival between two chemoradiotherapy (CRT) regimens for inoperable non-metastatic oesophageal cancer patients. This secondary analysis aimed to describe the impact of CRT on health-related quality of life (HRQOL), physical functioning, dysphagia, fatigue and pain and to evaluate whether baseline HRQOL domains can predict overall survival. PATIENTS AND METHODS: A total of 267 patients were randomly assigned to receive with 50 Gy of radiotherapy in 25 fractions six cycles of FOLFOX or four cycles of fluorouracil and cisplatin on day 1. HRQOL was prospectively assessed using the European Organization for Research and Treatment of Cancer Core Quality of Life Questionnaire version 3.0 with the oesophageal cancer module (QLQ-OES18). RESULTS: Both groups showed high baseline compliance. Subsequently, compliance reduced to 41% at the 6-month follow-up. Baseline HRQOL scores showed no statistical differences between treatment arms. During treatment, both groups exhibited lower physical and social functioning and increased fatigue and dyspnoea, although dysphagia moderately improved in the fluorouracil-cisplatin arm only (p = 0.047). During follow-up, HRQOL scores revealed no significant differences between chemotherapy regimens. Linear mixed model exhibited a treatment-by-time interaction effect for dysphagia (p = 0.017) with a greater decrease in dysphagia in the fluorouracil-cisplatin group. Time until definitive deterioration analysis showed no significant differences in global HRQOL, functional or main symptom domains. However, time until definitive deterioration was significantly longer for the fluorouracil and cisplatin arm compared with FOLFOX for appetite loss (p = 0.002), QLQ-OES-18 pain (p = 0.008), trouble swallowing saliva (p = 0.011) and trouble talking (p = 0.020). CONCLUSION: Analyses of HRQOL scores revealed no statistically significant differences between patients with inoperable non-metastatic oesophageal cancer treated by FOLFOX versus those treated with a fluorouracil-cisplatin regimen as part of definitive CRT.

Antibody targeting of claudin-1 as a potential colorectal cancer therapy.

BACKGROUND: Metastatic colorectal cancer (mCRC) is one of the major causes of cancer-related death. Despite the substantial progress in mCRC management, it remains important to identify new therapeutic options and biological markers for personalized medicine. Here, we investigated the expression of claudin-1 (CLDN1), a major tight junction transmembrane protein, in the different colorectal cancer (CRC) molecular subtypes and then assessed the anti-tumor effect of a new anti-CLDN1 monoclonal antibody (mAb). METHODS: Gene expression profiling and immunochemistry analysis of normal and tumor tissue samples from patients with stage IV CRC were used to determine CLDN1 gene expression. Then, the 6F6 mAb against CLDN1 extracellular part was generated. Its effect on CRC cell cycle, proliferation, survival and migration was assessed in vitro, using a 3D cell culture system, flow cytometry, clonogenic and migration assays. In vivo, 6 F6 mAb efficacy was evaluated in nude mice after subcutaneous xenografts or intrasplenic injection of CRC cells. RESULTS: Compared with normal mucosa where it was almost exclusively cytoplasmic, in CRC samples CLDN1 was overexpressed (p < 0.001) and mainly localized at the membrane. Moreover, it was differentially expressed in the various CRC molecular subtypes. The strongest expressions were found in the consensus molecular subtype CMS2 (p < 0.001), the transit-ampliflying (p < 0.001) and the C5 subtypes (p < 0.001). Lower CLDN1 expression predicted a better outcome in the molecular subtypes C3 and C5 (p = 0.012 and p = 0.004, respectively). CLDN1 targeting with the 6 F6 mAb led to reduction of survival, growth and migration of CLDN1-positive cells. In preclinical mouse models, the 6F6 mAb decreased tumor growth and liver metastasis formation. CONCLUSION: Our data indicate that CLDN1 targeting with an anti-CLDN1 mAb results in decreased growth and survival of CRC cells. This suggests that CLDN1 could be a new potential therapeutic target.

Molecular subtypes of metastatic colorectal cancer are associated with patient response to irinotecan-based therapies.

BACKGROUND: Currently, metastatic colorectal cancer is treated as a homogeneous disease and only RAS mutational status has been approved as a negative predictive factor in patients treated with cetuximab. The aim of this study was to evaluate if recently identified molecular subtypes of colon cancer are associated with response of metastatic patients to first-line therapy. PATIENTS AND METHODS: We collected and analysed 143 samples of human colorectal tumours with complete clinical annotations, including the response to treatment. Gene expression profiling was used to classify patients in three to six classes using four different molecular classifications. Correlations between molecular subtypes, response to treatment, progression-free and overall survival were analysed. RESULTS: We first demonstrated that the four previously described molecular classifications of colorectal cancer defined in non-metastatic patients also correctly classify stage IV patients. One of the classifications is strongly associated with response to FOLFIRI (P=0.003), but not to FOLFOX (P=0.911) and FOLFIRI + Bevacizumab (P=0.190). In particular, we identify a molecular subtype representing 28% of the patients that shows an exceptionally high response rate to FOLFIRI (87.5%). These patients have a two-fold longer overall survival (40.1 months) when treated with FOLFIRI, as first-line regimen, instead of FOLFOX (18.6 months). CONCLUSIONS: Our results demonstrate the interest of molecular classifications to develop tailored therapies for patients with metastatic colorectal cancer and a strong impact of the first-line regimen on the overall survival of some patients. This however remains to be confirmed in a large prospective clinical trial.

Impact of neoadjuvant chemotherapy cycles prior to interval surgery in patients with advanced epithelial ovarian cancer.

OBJECTIVES: Complete surgery with no macroscopic residual disease (RD) at primary (PDS) or interval debulking surgery (IDS) is the main objective of surgery in advanced epithelial ovarian cancer (EOC). The aim of this work was to evaluate the impact on survival of the number of neoadjuvant chemotherapy (NAC) cycles before IDS in EOC patients. METHODS: Data from EOC patients (stages IIIC-IV), operated on between 1995 and 2010 were consecutively recorded. NAC/IDS patients were analyzed according to the number of preoperative cycles (<4=group B1; >4=group B2) and compared with patients receiving PDS (group A). Patients with complete resection were specifically analyzed. RESULTS: 367 patients were analyzed, 220 received PDS and 147 had IDS/NAC. In group B, 37 patients received more than 4 NAC cycles (group B2). Group B2 patients presented more frequently stage IV disease at diagnosis (p<0.01) compared to groups A and B1. The rate of complete cytoreduction was higher in group B (p<0.001). Patients with no RD after IDS and who had received more than 4 NAC cycles had poor survival (p<0.001) despite complete removal of their tumor (relative risk of death after multivariate analysis of 3 (p<0.001)) with an independent impact from disease stage and WHO performance status. CONCLUSIONS: Patients with advanced EOC receiving complete IDS after more than 4cycles of NAC have poor prognosis. Despite worse prognostic factors observed in this group of patients, our study reinforces the concept of early and complete removal of all macroscopic tumors in the therapeutic sequence of EOC.

SAE: an R package for early stopping rules in clinical trials.

In the case of an unexpected high frequency of serious adverse events (SAE), statistical methods are needed to help in the decision making process as to continuation of accrual to the trial. This paper describes an R package, named SAE that implements a method recently developed by defining stopping rules after each observed SAE. The package function control for excessive toxicity either during the trial at the observation of each SAE (function SAE) or during the planning phase of a clinical trial (function DESIGN). This description and the package documentation are complementary to help the users to apply the method. The main difficulty in the implementation of the method is the choice of a priori parameters. Data from an ongoing clinical trial are presented as an example to improve the understanding and the use of the package.

A large-scale candidate gene approach identifies SNPs in SOD2 and IL13 as predictive markers of response to preoperative chemoradiation in rectal cancer.

Neoadjuvant radiochemotherapy followed by total mesorectal excision is now the standard treatment for locally advanced rectal cancer. However, tumor response to chemoradiation varies widely among individuals and cannot be determined before the final pathologic evaluation. The aim of this study was to identify germline genetic markers that could predict sensitivity or resistance to preoperative radiochemotherapy (RT-CT) in rectal cancer. We evaluated the predictive value of 128 single-nucleotide polymorphisms (SNPs) in 71 patients preoperatively treated by RT-CT. The selected SNPs were distributed over 76 genes that are involved in various cellular processes such as DNA repair, apoptosis, proliferation or immune response. The SNPs superoxide dismutase 2 (SOD2) rs4880 (P=0.005) and interleukin-13 (IL13) rs1800925 (P=0.0008) were significantly associated with tumor response to chemoradiation. These results reinforce the idea of using germline polymorphisms for personalized treatment.

Combined cetuximab and trastuzumab are superior to gemcitabine in the treatment of human pancreatic carcinoma xenografts.

BACKGROUND: Pancreatic carcinoma remains a treatment-refractory cancer with a poor prognosis. Here, we compared anti-epidermal growth factor receptor (EGFR) and anti-HER2 monoclonal antibodies (2mAbs) injections with standard gemcitabine treatment on human pancreatic carcinoma xenografts. MATERIALS AND METHODS: Nude mice, bearing human pancreatic carcinoma xenografts, were treated with either combined anti-EGFR (cetuximab) and anti-HER2 (trastuzumab) or gemcitabine, and tumor growth was observed. RESULTS AND CONCLUSION: In first-line therapy, mice survival was significantly longer in the 2mAbs group compared with gemcitabine (P < 0.0001 for BxPC-3, P = 0.0679 for MiaPaCa-2 and P = 0.0019 for Capan-1) and with controls (P < 0.0001). In second-line therapy, tumor regressions were observed after replacing gemcitabine by 2mAbs treatment, resulting in significantly longer animal survival compared with mice receiving continuous gemcitabine injections (P = 0.008 for BxPC-3, P = 0.05 for MiaPaCa-2 and P < 0.001 for Capan-1). Therapeutic benefit of 2mAbs was observed despite K-Ras mutation. Interestingly, concerning the mechanism of action, coinjection of F(ab')(2) fragments from 2mAbs induced significant tumor growth inhibition, compared with controls (P = 0.001), indicating that the 2mAbs had an Fc fragment-independent direct action on tumor cells. This preclinical study demonstrated a significant improvement of survival and tumor regression in mice treated with anti-EGFR/anti-HER2 2mAbs in first- and second-line treatments, compared with gemcitabine, independently of the K-Ras status.

Early stopping rules in clinical trials based on sequential monitoring of serious adverse events.

Several multistage or group sequential statistical methods have been developed for defining stopping rules in terms of efficacy in phase II and III clinical trials, but they rely on interim analyses after the inclusion of a fixed number of patients. These methods, however, need to be adapted for the evaluation of serious adverse events (SAEs), which can occur relatively early in the trial. A high frequency of their occurrence may require the trial to close early. The methods developed here define stopping rules after the occurrence of each SAE by comparing the number of patients included to the number of patients satisfying maximum SAE criteria. The nominal type I error, power, and average sample number (ASN) under specific hypotheses are obtained through simulations. Data from a clinical trial are presented as an example.

Two-part statistics with paired data.

In epidemiology, we often study data from a mixed distribution, i.e. with a clump of observations at zero and positive continuous data. Lachenbruch developed statistics for this kind of data for independent samples. These tests are the sum of one test for equality of proportions of zero values and one conditional test for the continuous distribution. This paper concerns the adaptation of these tests to paired samples. Like Lachenbruch, we developed two statistics, which tend to a two-degree-of-freedom chi2 distribution. These two-part statistics are the sum of McNemar's test for testing the equality of proportions of zero values, and the Wilcoxon signed-rank test or the paired Student's test for testing the equality of the distribution of positive values. We studied the behaviour of these tests for various proportions of zeros, and mean values of the continuous distribution. All tests are efficient when the smaller proportion of zero values corresponds to the population with the larger mean. In all other situations, the two-part statistics are superior to the others. These methods are applied to a matched case-control study of lower limb venous insufficiency.