home > ict > spring 2012 > data is king
International Clinical Trials

Data is King


With comprehensive information on endpoints, timelines, populations and safety results readily available as a benchmark, the FDA approval process can offer cancer pivotal trials real flexibility. Less than one in five cancer compounds entering clinical trials will ever be approved for use in humans (1). Making the problem worse is the fact that, in relation to other therapeutic categories, oncology drugs are proportionately more likely to fail in expensive, late-stage trials rather than earlier in the development process (2).

Furthermore, while no area of drug development can be considered straightforward, cancer is certainly more complicated than any other major category of current research. A plethora of potential endpoints, numerous sub-populations (including a growing reliance on genetic testing), indications that cover a range of potential linesof- therapy and combination regimens, are just a few of the challenges that compound the difficulty of bringing cancer drugs to market. Of course, these complications must be considered alongside the issues that are common to any clinical development programme, such as trial design, sizing, site placement, forecasting the time and resources required, understanding the optimal point at which to seek approval, and predicting how regulatory agencies will react despite the fact that this is rarely known in advance with any degree of certainty.

This article includes high level findings from analysis of the insiteinvestigator™ database which currently contains comprehensive information on design, endpoints, populations, size, placement, timelines, efficacy outcomes and safety results for every pivotal trial supporting a cancer drug approved by the US FDA from 2005 through 2011. By providing benchmarks for successful trials, the database provides a tool to help decision-makers separate potential winners from probable losers at an early stage and to prevent errors in planning, whether in trial design or execution, from contributing to regulatory rejection. The article covers cancer drug approval trends, approval requirements, pivotal trial design, size, location and key timing elements such as recruitment periods and patient accrual per site within specific cancer types.

Approval Requirements

As a rule, the FDA recommends that applications include data from “at least two adequate and well-controlled clinical trials” as a condition of standard drug approval (3). In cancer, however, the high degree of unmet medical need and the urgency of offering new treatments for potentially fatal diseases mean that approvals are based on a wide range of clinical data packages.




This is demonstrated in Table 2, which shows that the “two adequate and well-controlled trials” standard is actually an unusual occurrence in cancer, as only two per cent of all indications received approval using this approach. Indeed, approved applications have been based on pivotal trials ranging from as little as a single Phase 1 pharmacokinetic study (asparaginase Erwinia chrysanthemi, approved in 2011 as part of a combination regimen for acute lymphoblastic leukaemia) to as much as three separate Phase 3 trials (rituximab’s 2006 approval for use with CHOP chemotherapy in newly diagnosed non-Hodgkin l.ymphoma patients).

Despite the variance, certain thresholds are evident. For example, 75 per cent of all applications and 56 per cent of orphan indications were directly supported with at least one Phase 3 trial. Standards are predictably higher for non-orphan indications, where 88 per cent of applications included a Phase 3 trial. Of indications approved without Phase 3 support, most were for patients who had failed or who demonstrated intolerance to earlier lines of therapy, which is understandable given the limited treatment options available to these groups.

In addition to demonstrating the range of required data, Table 2 also reveals that 59 per cent of all indications were approved with a single Phase 3 trial. Nearly three-quarters of all non-orphan drugs fall in this category, and it is the most common scenario for orphan drugs as well. In the absence of specific guidance from regulatory agencies, this category may serve as the best available proxy for establishing the scope of a clinical trial programme necessary to gain approval.

Pivotal Trial Design

The database also captures elements of pivotal trial design. Using comparisons between solid tumours and haematological malignancies, the following explore several of the more important design features: primary endpoints, number of arms and control types. Primary Endpoints Overall survival, defined as the time from randomisation until death from any cause, is the FDA’s gold standard for drug approval in the cancer field. From a development standpoint, however, it is the most difficult and costly endpoint to demonstrate as it requires higher patient populations and longer follow-up periods. As we have seen in the preceding section, the FDA is often willing to accept less than gold standard evidence when doing so will make new therapies available if the data is sufficient to show that these new therapies are both safe and effective.

Indeed, as shown in Table 3, overall survival served as the primary endpoint in only 18 per cent of all pivotal trials during our time period. Although it was the most common primary endpoint in solid tumours, it still represented just one third of such trials, barely edging out progression-free survival, which was the primary endpoint in 29 per cent of solid tumour pivotal trials. More striking is its near-total absence in haematologic malignancies, where only four per cent of pivotal trials featured overall survival as a primary endpoint. In these cancers, response rate and the closely related duration of response/clinical benefit are the predominant primary endpoints, having been used to support 59 per cent of all approvals.


Number of Arms

Figure 2 shows the frequency of multiple and single arm trials in cancer pivotal trials. Two-arm trials dominate the solid tumour category, with 84 per cent of all studies falling into this category and the remainder being split equally between single and double arm trials. The picture is quite different in haematologic malignancies, however, where single arm trials account for 49 per cent of all haematologic pivotal studies and are as common in these cancers as two arm trials. It is important to note that the data in Figure 2 is based on individual pivotal trials and, as we have seen, indications are often approved on the basis of more than one trial. However, there were eight approvals in our time period that were made on the basis of a sole, single arm trial. Most were for haematologic cancers, and these applications tended to be for conditions with small populations or for use in second line or later patients.

Also of interest is the five per cent of trials in the dataset that featured three arms, all but one of which were in the solid tumour category. Three arms were used to test different doses of the investigational drug or in cases where the agent was included in both monotherapy and combination regimens. In each of these instances, a single arm received approval – a fact that suggests more extensive testing at an earlier phase might have allowed for cost savings through the elimination of the third arm. This must be weighed against the additional time required to conduct further research before moving into the pivotal study. However, three-arm trials also slow the process by diverting scarce patients into an arm that will ultimately be discarded. Data on recruitment times and accrual rates, discussed later in this article, may assist planners in evaluating the time trade-off between conducting additional small studies or moving directly into the pivotal trial with a three arm study.

Control Types

Selecting a comparison arm for cancer trials is generally more challenging than in other types of drug development. On the one hand, there is a desire to demonstrate the treatment’s effect as clearly as possible, something that is best done with a placebo control. On the other hand, review boards are often reluctant to approve these types of trials, as there are serious ethical issues involved in featuring a placebo arm with patients suffering from a potentially fatal disease. Instead, common practice calls for using an active control (usually the existing standard-ofcare regimen) in the control arm with the investigational agent used either in combination with the standard-of-care or, in cases where more data is available, as monotherapy.

In Figure 3 we see how these general rules are applied in practice. If we ignore single arm trials in haematologic malignancies for the moment, active controls do in fact serve as the most common type of comparison arm for both cancer types. There are also, however, a fair number of placebo-controlled pivotal trials, particularly in the solid tumour area. Usually, these were for patients who had failed earlier lines of therapy, but this was not always the case as two trials (both in renal cell carcinoma) achieved first-line indications via placebo-controlled trials. Placebo-controlled trials were substantially more likely to be used in monotherapy (82 per cent) than in combination therapy trials (18 per cent). This is somewhat surprising given the aversion to giving cancer patients placebo-only treatment, but may, at least in part, be reflective of the fact that there are no existing standards-of-care for certain advanced cancers.

Trial Performance Metrics

In addition to understanding key design characteristics, planners also need to know how their trials are likely to perform in the implementation stage. Several key findings can be drawn from the database regarding time from pivotal trial initiation to approval, accrual rates and trial placement patterns.

Development Time

Figure 4 shows the median time required from pivotal trial initiation (defined here as the date of first patient enrolment) to FDA approval by cancer type (several cancers are excluded because of limited data). For the cancers shown, the overall median time is approximately 63 months or a little over five years. However, this ranges considerably from a low of 36 months for kidney cancer to a high of more than 100 months in skin cancer trials. The analysis suggests that these variations are driven by a combination of several factors including the size of the available population, the number of competitive trials taking place at the same time, and how the existing standard-of-care is viewed.

Accrual Rates

Several of the factors that slow development times are beyond the planner’s control. Subject to constraints on the budget and site availability, however, it is generally possible to speed the process by increasing the number of locations hosting the trial. The information in Figure 5 (see page 34) is of interest for this reason as it shows the median accrual rate for various types of cancer. (As with the time to approval data, some cancer types are not shown.) Note that accrual rate is defined as the number of patients recruited per site per month and the number of total months is based on the period from first patient enrolled to last patient enrolled.

The median accrual rate for the included conditions is 0.23 patients per site per month. Using this information, a hypothetical trial with a recruitment goal of 1,000 patients could expect to achieve full enrolment in 43.5 months with 100 sites and 34.8 months with 125 sites. Of course, there is no such thing as a ‘median cancer’ and, as we have seen with the time to approval data, recruitment rates vary considerably between conditions. Unsurprisingly, those with the fastest times to approval also tend to have the highest accrual rates.


Pivotal Trial Location

A great deal of attention has been devoted to the globalisation of clinical trials in recent years, particularly as this relates to placement in developing countries. In that context, the information in Figure 6 is interesting as it shows that, at least among the pivotal trials supporting cancer drug approval in our time period, three quarters of all sites were in North America or western Europe while China and India – locations that are at the heart of the debate over globalisation – had only two per cent of all sites combined. Considering that trials in emerging countries are generally able to recruit patients more rapidly than is the case in more traditional locations, an argument can be made that there is too little rather than too much clinical research taking place in emerging countries. This is especially relevant given what we have seen about accrual rates.



Analysis of the database reveals a number of valuable facts about cancer pivotal trials. Most notably, we have seen that these studies are more varied and the FDA approval process is more flexible than may be commonly understood. For example, in comparison to other drug types, it is relatively rare for an oncology agent to be approved based on two or more Phase 3 trials and overall survival, the preferred primary endpoint, is used in only a minority of all cases, and almost never in haematologic malignancies.

Further, while it is known that these trials take a considerable amount of time, we now have specific measurements to assist in the forecasting process and can see how changing the number of sites is likely to affect the accrual period. Finally, the very limited level of trial placement in emerging locations suggests that there is an opportunity to speed the development cycle while still restricting participation to the highest quality institutions in these countries.


  1. 1. DiMasi J et al, Trends in risks associated with new drug development: success rates for investigational drugs, Clinical Pharmacology & Therapeutics, March 2010
  2. 2. DiMasi J and Grabowski H, Economics of new oncology drug development, Journal of Clinical Oncology 25(2): pp209-216, 2007
  3. 3. Guidance for Industry: Clinical Trial Endpoints for the Approval of Cancer Drugs and Biologics, Food and Drug Administration, May 2007

Read full article from PDF >>

Rate this article You must be a member of the site to make a vote.  
Average rating:

There are no comments in regards to this article.


Todd Clark is the President of Value of Insight Consulting. Over the past 20 years he has worked with leading drug and biotech companies, advising them on global issues regarding clinical trial design, life cycle management, regulatory compliance, marketing strategy and more. He is the author of numerous industry publications including PharmaHandbook: A Guide to the International Pharmaceutical Industry and, most recently, Oncology Clinical Trials: The Roadmap to FDA Approval, an in-depth analysis of information from the insiteinvestigator™ database, and holds an MBA.

Todd Clark
Print this page
Send to a friend
Privacy statement
News and Press Releases

Owen Mumford Launches New 2-Step Single-Use Auto-Injector Platform Aidaptus®

OXFORD, England, 22nd September - Owen Mumford Pharmaceutical Services, a division of Owen Mumford Ltd. has launched its new Aidaptus® auto-injector platform following successful completion of development. Aidaptus is a 2-step, spring-powered, single use auto-injector with a versatile design that accommodates both 1mL and 2.25 mL prefilled glass syringes in the same base device. It also features stopper sensing technology with a self-adjusting plunger that automatically adapts to the individual stopper positions and different fill volumes in each syringe, with no change parts required. The auto-injector is available with two different spring strengths to accommodate a variety of drug viscosities.
More info >>

White Papers

Delivering Fit For Purpose Biomanufacturing CHO Cell Lines

Fujifilm Diosynth Biotechnologies

The challenge during mammalian cell line development is to identify and isolate stable, high expressing cell lines producing product with the appropriate critical product quality attributes rapidly, reproducibly and with relative ease. Obtaining a host cell line that inherently exhibits desirable biomanufacturing attributes can therefore have a significantly positive effect on the identification of recombinant cell lines with desired traits during cell line development screens. In this study, we demonstrate that it is possible to exploit intrinsic heterogeneity within host cell populations and identify host cell lines which are more “fit for purpose”.
More info >>




©2000-2011 Samedan Ltd.
Add to favourites

Print this page

Send to a friend
Privacy statement