摘要
适应性设计由于其灵活可变的优势一直是生物统计学家关注和研究的热点。特别是近年来,随着计算机和网络技术的发展,阻碍适应性设计应用的诸多障碍得以解除,其在新药临床试验中开始得到广泛应用,其优势亦日益得到各国药政机构、申报者和试验研究者等的重视。交互式语音应答系统、电子数据捕获、临床试验模拟预测评价被称为促进适应性设计在临床试验中得以应用的三大关键技术。交互式语音应答系统和基于Web的中央随机化系统可以利用电话和Internet网络即时通知研究者试验药物的随机编号,为适应性设计中调整随机分配方案提供可能性。电子数据捕获平台的应用不仅可以缩短试验时间,及时获得高质量的试验数据,从而满足适应性设计的要求,而且能够降低试验成本,提高试验效率,保证整个试验的可靠性。而临床试验模拟预测评价系统可以在适应性临床试验开始前即对整个试验流程以及可能出现的想定进行计算机模拟,并对可能出现的结果进行预测和评价以指导研究者和生物统计人员选择最优的方案设计,以达到优化试验设计、提高试验成功把握度的目的。
EDC系统是促进适应性设计在新药临床试验中的关键技术,它的优劣直接关系到适应性临床试验的实施和质量。而EDC系统构建的难点在于如何解决e-CRF设计的可复用性问题,即在设计不同临床试验病例报告表时,如何充分利用它们之间的共性以提高e-CRF设计的效率。此外,适应性设计在临床试验中的应用也要求临床试验模拟预测评价系统的研究和开发,而该系统得以构建开发的基础在于模拟预测评价方法的研究。基于以上考虑,本课题适应性临床试验中需要的EDC系统构建策略和两阶段适应性设计的模拟预测评价方法展开研究。具体研究内容如下:
(1)基于结构化病例报告表的EDC系统构建策略
系统分析我国新药临床试验中病例报告表结构化的特点。并根据这一特点,分别建立临床试验通用数据库模型和e-CRF元数据库以解决e-CRF设计的可复用性问题。采用EAV数据结构和传统关系型数据结构联合构建的通用数据库模型不依赖于临床试验收集指标的变化而变化,具有通用性的特点。这大大简化了每个临床试验中e-CRF数据库的建立,解决了EDC系统数据库的可复用性。而由XML文档、XSL模板文档和JavaScript核查文件共同组成的e-CRF元数据库,记录了临床试验指标的数据类型、Web页面表现格式、逻辑核查条件、分类变量取值选项以及指标注释等相关属性。利用e-CRF元数据库搭建整个病例报告表的XML文档和XSL文档可以较为方便实现病例报告表Web录入页面的自动生成。
此外,在EDC系统构建中,XML语言不仅用来建立e-CRF元数据库,而且在e-CRF与EDC系统数据库之间作为中间载体实现数据的存储与读取。这一方面可以方便实现EDC系统与中央随机化、不良事件的监测与报告等其它系统之间的数据传输,增强系统本身的兼容性,为不同系统之间的整合和集成一体化工作奠定基础;另一方面,以XML数据文件作为数据传输的中间载体,可以重复利用Web页面XSL文档,提高代码使用效率,做到“一处编写,多处重用”。
(2)两阶段适应性设计模拟预测评价方法研究
以二分类资料的两阶段适应性双臂试验为例,在试验方案的设计阶段即采用Monte Carlo模拟方法,对试验中可能出现的多种想定以及所做出的适应性调整进行计算机模拟,并对可能出现的结果进行预测评价以选出最优的试验参数α1、α0、α2和第二阶段样本量方案。这种模拟预测评价方法允许在期中分析时根据已有数据信息调整第二阶段样本量,提高试验的灵活性,而且在试验方案中即对可能做出的样本量调整进行预计划,更大程度维持了试验的完整性,提高了试验质量。并根据新药临床试验的实际,考虑可能出现的最大样本量限制、两组不等样本量、终点变量为正态分布或生存数据等其它分布类型时,对两阶段适应性设计的模拟预测评价方法进行进一步拓展。
此外,模拟比较了两阶段适应性设计中多种全面分析方法,发现两阶段合并数据分析的方法不仅可以保持较高的总检验效能,而且通过采用加法原则能够对总I类错误较好的控制。而在两阶段p值合并分析的几种方法中,INM法和MSP法虽然总检验效能明显高于Bauer-K?hne法,但总I类错误不能得到很好的控制,出现明显的alpha膨胀现象。而Bauer-K?hne法虽然在总检验效能方面存在劣势,但总I类错误仅略高于0.05,在可接受范围内。
本文通过对基于结构化病例报告表的EDC构建策略和两阶段适应性设计的模拟预测评价方法,为EDC平台和临床试验模拟预测评价系统的构建打下理论和算法基础,以进一步促进适应性临床试验在我国的应用。
Adaptive design has been attached much importance to by biostatisticians in their studies due to its flexibility. In the past few years, with the development of computer and network technology, the obstacles which block the application of adaptive design were removed and its advantages have been paid more and more attention by regulatory agencies, sponsors and investigators. IVRS, EDC and clinical trial simulation, forecast & evaluation system have been regarded as three key modules of promoting application of adaptive design. IVRS and IWRS, which can inform the investigators the random number of test drug instantly, offer the possibility of adjust the random allocation scheme in adaptive design. The application of EDC, can not only cut down the trial time and obtain high quality data in time, but reduce the costs, improve the efficiency and ensure its reliability. In clinical trial simulation, forecast & evaluation system, the whole trial process and conceivable scenarios can be simulated by computer simulation and the possible results will be forecasted and evaluated to guide the investigators and biostatisticians to optimize the trial design and improve the power of clinical trial.
EDC, which has the direct relationship with the implementation and quality of adaptive clinical trial, is the key factor of promoting the application of adaptive design in clinical trials. The difficulty of EDC building is the reusability of e-CRF design, which is to make use of the commonness of case report forms as much as possible to improve the efficiency when designing different e-CRFs. Moreover, the system of simulation, forecast & evaluation has to be paid much attention to due to the application of adaptive design. In the system of simulation, forecast & evaluation, the whole trial flow can be simulated and the corresponding results will be forecasted and evaluated to help the investigators and biostatisticians choose the optimal trial design. However, the basis of this system is the study of simulation, forecast & evaluation method. Considering what we have discussed above, this study focuses on the methods of EDC building,simulation, forecast and evaluation method for two stage adaptive design. The main works are introduced as follows:
(I)Study of EDC building based on structural case repot form
According to the structural characteristic of case report form, which is investigated systematically based on the clinical trials in our country, general database model and e-CRF metadata repository have been built to deal with the problem of reusability. The structure of general database model, which is constructed by using EAV structure and traditional relational structure, does not change with the change of indexes in different clinical trials. It simplifies the process of e-CRF database building for each clinical trial and solves the reusability of EDC database. The data type, format in Web page, check condition, option for categorical variable, label, etc are include in e-CRF metadata repository, which is composed of XML file, XSL file and JavaScript check file. Due to e-CRF metadata repository, the XML file and XSL file for the whole case report form can be written easily to build the Web page of e-CRF for data entry.
Furthermore, XML is not only to construct e-CRF metadata repository, but a medium between e-CRF and EDC database for storing and reading. It can help the data transportation between EDC and other related systems in clinical trials. Besides, in virtue of XML, the XSL file can be used for Web page generation repeatedly and the efficiency of programming is improved.
(II)Study of simulation, forecast & evaluation method for two-stage adaptive design
Taken adaptive two-stage double-arm clinical trial for dichotomous variables as an example, Monte Carlo simulation is performed at the stage of protocol design to simulate multiple conceivable scenarios and adaptations. Then the corresponding results could be forecasted and evaluated to determine the optimal parametersα1、α0、α2and sample size of stage II. This method, in which the sample size of stage II can be adapted based on the accrued data at the interim stage, enhances the flexibility of the clinical trial. Besides, the integrity of the trial can also be ensured much more by preplanning the possible sample size adaptation in protocol design. Moreover, this method is generalized when sample size constraint, unequal sample size and different distribution for endpoint variables are considered.
The methods of overall analysis are also compared by Monte Carlo simulation. It has been found that the method of pooled data analysis have a higher overall test power and better constraint on overall type I error due to additive rule. Among the methods of combining p-values, although the overall test powers of INM method and MSP method are higher than Bauer-K?hne method, their overall type I error are not controlled finely. However, the overall type I error of Bauer-K?hne method, which has its inferiority in overall test power, is accepted in adaptive two-stage design
In this paper, the studies of EDC building method based on structural case report form and simulation, forecast & evaluation method for adaptive two-stage design provide theoretic and arithmetic basis for the building of EDC and simulation, forecast & evaluation system, which will further promote the application of adaptive clinical trials in our country.
引文
[1] Miller E, Murphy J. Implementing and Managing Adaptive Designs for Clinical trials. http://www.almacgroup.com/papers/Papers/Adaptive_ designs_article.pdf, 2005.
[2] Cai H, Xia J, Xu D, et al. A generic minimization random allocation and blinding system on web. Journal of Biomedical Informatics, 2006, 39: 706-719.
[3]蔡宏伟,夏结来,徐德忠,等.基于网络的最小化随机分组系统.第四军医大学学报, 2004, 25(14): 1278-1280.
[4]蔡宏伟,曹晓曼,夏结来.最小随机化分组系统在多中心临床试验中的应用.中国新药杂志, 2008, 17(14): 1264-1267.
[5] Food and Drug Administration. 21CFR PART11. Federal register. http://www.21cfrpart1.com/files/library/compliance/isp-riskbasedapproach. pdf, 2004.
[6] Food and Drug Administration. Guidance for computerized systems used in clinical trials. http://www.fda.gov/ora/compliance.ref/bimo/ffinalcct.pdf, 1999.
[7]孙亚林,贺佳,曹阳.国内外临床数据管理系统发展现状.第二军医大学学报, 2006, 27(7): 721-725.
[8] Sahoo U, Bhatt A. Electronic Data Capture (EDC)– A New Mantra for Clinical Trials. Quality Assurance. 2003, 10: 117-121.
[9] Bleicher P. Clinical trial technology: at the inflection point. BIOSILICO. 2003,1(5): 163-168.
[10] Lampe A.J. and Weiler J.M. Data capture from the sponsors’andinvestigators’perspectives: balancing quality, speed, and cost. Drug Information Journal. 1998, 32: 871–886.
[11] Jacobson A. The Emergence and Adoption of Electronic Data Capture(EDC) in Clinical Trials Data Collection. EDC today, Issue No. 1, http://www.edcmanagement.com..
[12] Hyde A.W. The Changing Face of Electronic Data Capture: From Remote Data Entry to Direct Data Capture. Drug Information Journal. 1998, 32:1089-1092.
[13] Yamamoto K, Matsumoto S, Tada H, et al. A Data Capture System for Outcomes Studies that Integrates with Electronic Health Records: Development and Potential Uses. Journal of Medical Systems, 2008, 32(5): 423-427.
[14] Clinical Trial EDC Task Group, PhRMA Biostatistics & Data Management Technical Group. Electronic Clinical Data Capture (position paper revision1). May 1st, 2005.
[15]卜擎燕,熊宁宁,邹建东,等.电子数据获取:实现更加优质与高效的临床研究.中国临床药理与治疗学, 2007, 12(4): 455-459.
[16] Susanne Prokcha. Practical Guide to Clinical Data Management(2nd Edition). Taylor & Francis Group, 2006: 43-73.
[17] Dimen?s E, Johansson D, Palmblad M, et al. Clinical Operation Online (COOL)-A World Wide Web-Based Approach to Running Clinical Trials: Results from Two International Multicenter Gastrointestinal Trials. Drug Information Journal, 2001, 35: 745-753.
[18] Bushnell D.M., Martin M.L., Parasuraman B. Electronic Versus Paper Questionnaires: A Further Comparison in Persons with Asthma. Journal of Asthma, 2003, 40(7): 751-762.
[19] Ronald W. Helms. Data Quality Issues in Electronic Data Capture. Drug Information Journal, 2001, 35: 827-837.
[20] Spink C. Pharmaceutical Clinical Development-Electronic Data Capture (EDC) as a means for e-clinical trial success. IBM Global Services– LifeSciences; 2002.
[21]李伟,熊宁宁,卜擎燕,等.电子数据获取系统在临床研究质量控制中的应用优势讨论.中国新药杂志, 2009, 18(2): 1081-1086.
[22] Eisenstein E.L., Collins R, Cracknell B.S., et al. Sensible approaches for reducing clinical trial costs. Clinical Trials, 2008, 5: 75-84.
[23] Pavlovi? I, Kern T, Miklav?i? D. Comparison of paper-based and electronic data collection process in clinical trials: Costs simulation study. Contemporary Clinical Trials, 2009, 30: 300-316.
[24] Oana A. Wlechers. The Move to EDC. Applied Clinical Trials, 2002, 11: 38-40.
[25] Glen de Vries. The Role of EDC in Adaptive Clinical Trials. Applied Clinical Trials, 2007, 3: 24-26.
[26] Bunn G. Scaling up EDC. Applied Clinical Trials, 2002, 8: 12-14.
[27] Welker J.A. Implementation of electronic data capture system: Barriers and solutions. Contemporary Clinical Trials, 2007, 28: 329-336.
[28] Smith S, Sambasivam S. Web Based Data Capture for Clinical Research. Proceedings of the Informing Science & IT Education Conference, 2008: 47-68.
[29]左伟明.即用即查XML数据标记语言参考手册.北京:人民邮电出版社, 2007: 3-5.
[30] Wübbelt P, Fermandez G, Heymer J. Clincial Trial Management and Remote Data Entry on the Internet based on XML Case Report Forms.Studies in Health technology and Informatics, 2000, 7: 333-337.
[31] Anne J. Buy vs. Build: Exploring EDC Technologies. EDC today, Issue No. 4, http://www.edcdatamanagement.com.
[32] Collins M, Ross E, Meropol N.J, et al. Using Metadata to Generate Web-Based Electronic Data Capture Forms. AMIA 2006 Symposium Proceedings, 2006: 896.
[33] Christopher J, Carol B, Mathew B, et al. A metadata-driven framework for generating field data entry interface in ecology. Ecological Informatics, 2007, 2: 270-278.
[34] Harris P.A, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)-A metadata-driven methodology and workflow process for providing translational research informatics support. Journal of Biomedical Informatics, 2009, 42: 377-381.
[35] Brandt CA, Morse R, Mattews K, et al. Metadata-driven creation of data marts from a EAV-modeled clinical research database. International Journal of Medical Informatics, 2002, 65(3): 225-241.
[36] Nadkarni PM, Brandt CM, Marenco L. WebEAV: Automatic Metadata-driven Generation of Web Interface to Entity-Attribute-Value Databases. Journal of the American Medical Informatics Association, 2000, 7(4): 343-356.
[37] Huff M, Hang DJ, Steven LE, et al. HELP the next generation: a new client-server architecture. Proceedings of 18th Symposium Computational Application of Medical Care, 1994: 271-275.
[38] 3M Health Information Systems. 3M Clinical Data Repository. Murray, Utah: 3M Corporation, 1998.
[39] Friedman C, Hripcsak G, Johnson S, et al. A generalized relational schemafor an integrated clinical patient database. Proceedings of 14th Symposium Computational Application of Medical Care, 1990: 335-9.
[40]蔡宏伟.基于网络的最小随机化分组设盲系统.第四军医大学,博士论文.
[41] Anh?j J. Generic Design of Web-Based Clinical Databases. Journal of Medical Internet Research, 2003, 5(4): e27. http://www.jmir.org/2003/ 4/e27/.
[42]孔玉,陆健,贺佳,等. CmpareDataQuick临床数据管理系统的开发与利用.中国卫生统计, 2007, 24(1): 80-81.
[43] Qing-Bin Gao, Yu Kong, Zheng Fu, et al. EZ-Entry: a clinical data management system. Computers in Biology and Medicine, 2008, 38(9): 1042-4.
[44] Anne J. Understanding Regulations, Guidelines, and Predicate Rules. EDC in Depth, Issue No.3, http://www.edcdatamanagement.com.
[45] ICH M2 EWG. Electronic Common Technical Document Specification. http://estri.ich.org/ectd/eCTD_Specification_v3_2_2.pdf, 2008.
[46] Souza T, Kush R, Evans JP. Global clinical data interchange standards are here! Drug Discovery Today, 2007, 12: 174-181.
[47] Kirk M, Karl M. Coding and Dictionaries. EDC today, Issue No.30, http://www.edcdatamanagement.com.
[48]卜擎燕,熊宁宁,邹建东,等. ICH国际医学用语词典(MedDRA):药事管理的标准医学术语集.中国临床药理学与治疗学, 2007, 12(5): 586-590.
[49] Wassmer G, Vandemeulebroecke M. A Brief Review on Software Developments for Group Sequential and Adaptive Designs. Biometrical Journal, 2006, 48(4): 732-7.
[50] Gallo P, Chuang-Stein C, Gragalin V, et al. Adaptive design in clinical drug development-an executive summary of the PhRMA Working Group (with discussion). Journal of Biopharmaceutical Statistics, 2006, 16: 275-83.
[51] Gould AL, Shih WJ. Modifying the design of ongoing trials without unblinding. Statistics in Medicine, 1998, 17: 89-100.
[52] Wittes J, Brittain E. The role of internal pilot studies in increasing the efficiency of clinical trials. Statistics in Medicine, 1990, 9: 65-72.
[53] Fisher L Self-designing clinical trials. Statistics in Medicine, 1998, 17: 1551-62.
[54] Bauer P, Brannath W. The advantages and disadvantages of adaptive designs for clinical trials. Drug Discovery Today, 2004, 9(8): 351-357.
[55] Chow SH, Chang M. Adaptive design methods in clinical trials. Boca Raton: Chapman & Hall/CRC; 2007: 11-20.
[56] Dragalin V. Adaptive Designs: Terminology and Classification. Drug Information Journal, 2006, 40: 425-435.
[57] Proschan MA. Sample size re-estimation in clinical trials. Biometrical Journal, 2009, 51(2): 348-357.
[58] Gould AL, Shih WJ. Sample size re-estimation without unblinding for normally distributed outcomes with unknown variance. Communications in Statistics– Theory and Methods. 1992, 21: 2833-2853.
[59] Birkeet MA, Day SJ. Internal pilot studies for estimating sample size. Statistics in Medicine, 1994, 13: 2455-2463.
[60] Bauer P, K?hne K. Evaluation of experiments with adaptive interim analysis. Biometrics, 1994, 50: 1029-41;
[61] Lehmacher W, Wassmer G. Adaptive sample size calculations in group sequential trials. Biometrics, 1999, 55: 1286-90.
[62] Proschan MA, Hunsberger SA. Designed extensions of studies based on conditional power. Biometrics, 1995, 51: 1315-24.
[63]于莉莉,薛富波,王素珍,等.临床试验适应性设计的反映变量-适应性随机化方法简介.中国卫生统计, 2008, 25(5): 534-538.
[64] Athreya KB, Karlin S. Embedding of urn schemes into continuous time Markov branching processes and related limit theorems. Annual of Mathematical Statistics, 1968, 39: 1801-7.
[65] Zelen M. Play the winner and the controlled clinical trial. Journal of the American Statistical Association, 1969, 64: 131-146.
[66] Wei LJ, Durham S. The randomized play-the-winner rule in medical trials. Journal of the American Statistical Association, 1978, 73: 840-3.
[67] Atkinson AC, Biswas A.. Adaptive biased-coin designs for skewing the allocation proportion in clinical trials with normal responses. Statistics in Medicine, 2005, 24: 2477-92.
[68] Chang M, Chow SC. A hybrid Bayesian adaptive design for dose response trials. Journal of Biopharmaceutical Statistics, 2005, 15: 677-691.
[69] Bretz F, Schmidli H, K?nig F, et al. Confirmatory Seamless Phase II/III Clinical Trials with Hypotheses Selection as Interim: General Concepts. Biometrical Journal, 2006, 48(4): 623-634.
[70] Schmidli H, Bretz F, Racine-Poon A. Bayesian predictive power for interim adaptation in seamless phase II/III trials where the endpoint is survival up to some specified timepoint. Statistics in Medicine, 2007, 26: 4925-38.
[71] Chow SC, Tu YH. On Two-Stage Seamless Adaptive Designs in Clinical Trials. Journal of the Formosan Medical Association, 2008, 107(12): 52-60.
[72] Chang M. Adaptive Design Theory and Implementation Using SAS and R.. Boca Raton: Chapman & Hall/CRC; 2008: 2-13.
[73] Lan KKG, DeMets DL. Discrete sequential boundaries for clinical trials. Biometrika, 1983, 70: 659-663.
[74] Pocock SJ. Group sequential methods in the design and analysis of clinical trials. Biometrika, 1977, 64: 191-9.
[75] O’Brein PC, Fleming TP. A Multiple Testing Procedures for Clinical Trials. Biometrics, 1979, 35: 549-556.
[76] Westfall PH, Tsai K, Ogenstad S, et al. Clinical Trial Simulation: a Statistical Approach. Journal of Biopharmaceutical Statistics, 2008, 18: 611-630.
[77] Bauer P. Multistage testing with adaptive designs. Biometrie und Informatik in Medizin und Biologie, 1989, 20: 130-148.
[78] Bauer P, Einfalt J. Application of Adaptive design-a Review. Biometrical Journal, 2006, 48(4): 493-506.
[79] Zaykin DV, Zhivotovsky LA, Westfall PH, Weir BS. Truncated product method for combining P-values. Genetic Epidemiology, 2002, 22: 170-85.
[80] Neuh?user M, Bretz F. Adaptive designs based on the truncated product method. BMC Medical Research Methodology, 2005, 5: 30.
[81] Cui L, Hung HMJ, Wang SJ. Modification of Sample Size in Group Sequential Clinical Trials. Biometrics, 1999, 55: 853-7.
[82] Chang M. Adaptive design method based on sum of p-values. Statistics in Medicine, 2007, 26: 2772-84.
[83] Wassmer G. A comparison of two methods for adaptive interim analysis in clinical trials. Biometrics, 1998, 54: 696-705.
[84] Metha CR, Bauer P, Posch M, et al. Repeated confidence interval for adaptive group sequential trials. Statistics in Medicine, 2007, 26: 5422-33.
[85] Wassmer G, Eisebitt R.“ADDPLAN: Adaptive Design-Plans and Analyses(Release 4.0). http://www.addplan.com, 2007.
[86] Cytel Inc.“East 5: Clinical Trial Design System.”http://www.cytel.com/ products/East/, 2007.
[87] Kairalla JA, Coffey CS, Muller KE. GLUMIP 2.0: SAS/IML Software for Planning Internal Pilots. Journal of Statistical Software, 2008, 28(7): 1-32.
[88]蒋志伟,夏结来,李婵娟,等.基于结构化病例报告表的EDC构建策略.中国新药杂志, 2009, 18(23): 2199-2204.
[89] CDISC CDASH Core and Domain Teams. Clinical data acquisition standards harmonization. http://www.cdisc.org, 2008.
[90]曹爱平,苏炳华,丁德云,等.临床试验中的病例报告表.中国卫生统计, 2000, 17(3): 188-191.
[91]李雪迎,康晓平,姚晨.药物临床试验安全性评价标准化数据结构的建立.中国临床药理学杂志, 2006, 22(2): 158-160.
[92] Oliveira AG, Salgado NC. Design aspects of a distributed clinical trials information system. Clinical Trials, 2006, 3: 385-396.
[93]李雪迎,康晓平,姚晨.水平型数据结构与垂直型数据结构占用存储空间的比较.中国卫生统计, 2006, 23(1): 39-42.
[94] Silva J, Wittes R. Role of clinical trial informatics in the NCI’s cancer informatics infrastructure. Proceeding of AMIA Symposium, 1999: 950-954.
[95] Silva J. Fighting cancer in the information age: an architecture for clinical scale clinical trials. MD Computing, 1999, 16(3): 43-44.
[96]胡镜清,吴圣贤,翟华强,等.病例报告表规范化与电子化研究现状及其中医药标准系统构建原则.中国中医药信息杂志, 2006, 13(9): 1-2.
[97] Simon R. Optimal two-stage designs for phase II clinical trial. Controlled Clinical Trials, 1989, 10: 1-10.
[98] Banerjee A, Tsiatis AA. Adaptive two-stage designs in phase II clinical trial. Statistics in Medicine, 2006, 25: 3382-95.
[99] Lin Y, Shih WJ. Adaptive two-stage designs for single-arm phase IIA cancer clinical trials. Biometrics, 2004, 60: 482-90.
[100] Emerson SS. Issues in the use of adaptive clinical trial design. Statistics in Medicine, 2006, 25: 3270-96.
[101] Bauer M, Bauer P, Budde M. A simulation program for adaptive two stage design. Computational Statistics & Data Analysis, 1998, 26: 351-371.