Interim analyses

Jacobson IM, Everson GT, Gordon SC, Kauffman R, McNair L, Muir A, McHutchison JG (2007) Interim analysis results from a phase 2 study of telaprevir with peginterferon alfa-2A and ribavirin in treatment-naive subjects with hepatitis C. AASLD 58th Annual Meet, Abstract 177 Johnson M (2006) Response to Atazanavir/ritonavir versus lopinavir/ritonavir equivalent or different efficacy profiles by HUl. AIDS 20 1987... 

Nolvadex Adjuvant Trial Organization (1983) Controlled trial of tamoxifen as adjuvant agent in management of early breast cancer interim analysis at four years by Nolvadex Adjuvant Trial Organization. Lancet i 257-261... 

Powles T, Eeles R, Ashley S (1998) Interim analysis of the incidence of breast cancer in the Royal Marsden Hospital tamoxifen randomizjsed chemoprevention trial. Lancet 352 98-101... 

Adang, A.E.P. and Hermkens, P.H.H., The contribution of combinatorial chemistry to lead generation an interim analysis, Curr. Med. Chem., 2001, 8, 985-998. 

De Angelo, D.J. et al.. Interim analysis of a phase If study of the safety and efficacy of gemtuzumab ozogamycin (Mylotarg) given in combination with cytarabine and daunorubicin in patients <60 years old with untreated acute myeloid leukemia, Proc. Am. Soc. Hem., 100,198a, Abstr. 745, 2002. 

Timing of any interim analysis Reason for sample size Power of clinical study Level of significance Criteria for termination of study Procedure for missing and imused data Reporting procedures for deviations from statistical plan... 

To illustrate how the type-I error may be elevated by using interim analysis we conduct a sampling experiment. The data below are 50 values of the Ritchie Index, a measurement of joint... 

Provision for making status reports, providing interim analysis reports and submitting the final report by specified dates should be made. 

In the bicalutamide prostate cancer programme, the adjuvant treatment of patients with advanced prostate carcinoma (T1-T4 NO/NX, MO) with bicalutamide (150 mg, once a day) was evaluated. 4052 patients were randomised to bicalutamide with best standard care (either radiation, prostatectomy or watchful waiting ), whereas 4061 patients received a placebo with best standard care. An initial reduction of prostate cancer recurrence, which was observed in an interim analysis, later was not confirmed [219]. As the survival time in the bicalutamide treatment group was decreased, the study was terminated ahead of time. 

Eckardt A, Wildfang I, Rades D, et al. Mulicenter phase II study of preoperative concurrent paclitaxel, carboplatin and radiotherapy in stage III/IV resectable cancer of the oropharynx and oral cavity an interim analysis (abstract 1658). Pro Am Soc Clin Oncol 2000 19 420a. 

Khuri FRL, Lee JJ, Winn RJ. Interim Analysis of Randomized Chemoprevention Trial of HNSCC. Proc Annu Meet Am Soc Clin Oncol 1999 18 A 1503. 

Pass H, Pogrebniak H, Steinberg SM, et al. Randomized trial of neoadjuvant therapy for lung cancer Interim analysis. Ann Thorac Surg 1992 53 992-1008. 

Granone P, Margaritora S, Cesario A, et al. Concurrent radio-chemotherapy in N2 non small cell lung cancer interim analysis. Eur J Cardio-thoracic Surgery 1997 12 366-371. 

Unfortunately, none of the seven randomized trials that have compared radiation therapy alone vs neoadjuvant cisplatin-containing chemotherapy plus radiation therapy demonstrated an improvement in overall or disease-free survival with combined-modality therapy (Table 2). Two studies actually demonstrated poorer survival with neoadjuvant chemotherapy. Souhami et al. (15) reported a significantly poorer survival rate with neoadjuvant chemotherapy in a small trial of patients with stage IIIB disease. This outcome was partly due to increased toxicity and poor compliance in patients who received chemotherapy. Another trial of neoadjuvant epirubicin and cisplatin was closed early when interim analysis revealed a significantly higher recurrence rate in the chemotherapy arm (16). These trials fail to provide any evidence that sequential cisplatin-containing chemotherapy and radiation therapy are of benefit. Possible explanations for the disappointing results include the effects of chemotoxicity, altered compliance, and possible accelerated repopulation of resistant clones after neoadjuvant chemotherapy. 

Citing the success of mitomycin-C in the treatment of anal cancer, a number of investigators have explored the use of this drug, usually in combination with fluorouracil, in patients treated with radiation for cervical and vulvar carcinomas (23-27). Roberts et al. (27) recently reported results of an interim analysis of a randomized trial, conducted in Venezuela, in which women with locally advanced cervical cancer were treated with radiation therapy with or without mitomycin. At the time of this interim analysis (published while the authors were continuing to accrue patients to the study), the authors noted a significant improvement in disease-free survival (p - 0.01) with chemotherapy but no significant difference in overall survival (p = 0.1). 

Note that this calculation must be undertaken on the blinded data to avoid any formal or informal treatment comparison. If such a comparison were to be made then there would be a price to pay in terms of the type I error. We will say much more about this in a later section dealing with interim analysis where the goal is to formally compare the treatment arms as the data accumulates. 

The considerations so far are based on the presumption that the type I error rate is divided equally across all of the comparisons. This does not always make sense and indeed it is not a requirement that it be done in this way. For example, with two comparisons there would be nothing to prevent having a 4 per cent type I error rate for one of the comparisons and a 1 per cent type I error rate for the other, providing this methodology is clearly set down in the protocol. We will see a setting below, that of interim analysis, where it is usually advantageous to divide up the error rate unequally. Outside of interim analysis, however, it is rare to see anything other than an equal subdivision. 

It is also possible to stop trials for reasons other than overwhelming efficacy, for example for futility, where at an interim stage it is clear that if the trial were to continue it would have little chance of giving a positive result. We will say more about interim analysis in a later chapter and in particular consider the practical application of these methods. 

Censoring in clinical trials usually occurs because the patient is still alive at the end of the period of follow-up. In the above example, if this were the only cause of censoring then all the censored observations would be equal to 24 months. There are, however, other ways in which censoring can occur, such as lost to follow-up or withdrawal. These can sometimes raise difficulties and we will return to discuss the issues in a later section. Also, at an interim analysis the period of follow-up for the patients still alive in the trial would be variable and this would produce a whole range of censored event times our methodology needs to be able to cope with this. 

A placebo-controlled randomised trial reported by Packer et al. (2001) investigated the effect of carvedilol on survival time in severe heart failure. Figure 13.1 shows the survival curve for each of the two treatment groups following the early termination of the trial at a planned interim analysis. 

One area that we briefly mentioned was interim analysis, where we are looking at the data in the trial as it accumulates. The method due to Pocock (1977) was discussed to control the type I error rate across the series of interim looks. The Pocock methodology divided up the 5 per cent type I error rate equally across the analyses. So, for example, for two interim looks and a final analysis, the significance level at each analysis is 0.022. For the O Brien and Fleming (1979) method most of the 5 per cent is left over for the final analysis, while the first analysis is at a very stringent level and the adjusted significance levels are 0.00052, 0.014 and 0.045. 

CH14 INTERIM ANALYSIS AND DATA MONITORING COMMITTEES... 

This method of calculating the conditional power assumes that the observed difference between the treatments at the interim stage is the true difference, termed the conditional power under the current trend. It is also possible to calculate conditional power under other assumptions, for example, that the true treatment difference in the remaining part of the trial following the interim analysis is equal to d. These calculations under different assumptions about how the future data should behave will provide a broad basis on which to make judgements about terminating the trial for futility. 

It is not a requirement that a trial must have an interim analysis, either for efficacy or for futility. In most long-term trials, however, where there is the opportunity for an interim evaluation then it may be something worth putting in place. The interim can involve only efficacy, only futility, or both and may indeed involve some other things as well, such as a re-evaluation of sample size (see Section 8.5.3). 

For practical reasons, however, the number of interims should be small in number. Undertaking interims adds cost to the trial and they also need to be very carefully managed. In particular, the results of each interim must be made available in a timely way in order for go/no-go decisions to be made in good time. Remember the trial does not stop to allow the interim to take place, recruitment and followup continues. It has been known in more than one trial for total recruitment to be complete before the results of the interim analysis become available - this is obviously a situation that you would want to avoid, the interim then becomes a... 

In line with this, the interim analysis plan was revised and only one interim was to be conducted for both efficacy and futility after 40 per cent of the patients had completed 3 months of follow-up. Since the two proposed analyses were not equally spaced, a spending functions were needed to revise the adjusted significance levels and these turned out to be 0.0007 and 0.0497. 

---