Trends in Oncology Trial Terminations: Industry Lessons Learned

For the drug development business, the probability of success in moving through the standard clinical trial phases to attain regulatory approval is low; these failures result in unrealized revenues for the biopharmaceutical companies who invested heavily in those drug programs. Pulling a candidate once it reaches proof of concept or later, is costly and the later it is pulled, the more expensive. We examine the reasons for industry-sponsored trial terminations across a 10-year period, and compare the early and later halves of the period to attempt to discern trends. We also identify the top company trial sponsors involved in these trials with an eye to performance trends.

For this analysis, Oncology terminated trials (excluding Supportive Care), sponsored by the biopharmaceutical industry, were assessed for the ten-year period from January 1, 2002 through December 31, 2011. Phase II, II/III and III trial data from the first five year period are compared to the more recent five years to reveal trends.

During the full ten year period, Citeline captured 1076 terminated trials [see Table 1]; 611 trials were initiated during 2002 – 2006 and 465 during 2007 through 2011.  The total numbers of Phase II, II/III and III trials started in these same periods were 2,884 and 3,025, respectively. Thus, for the 10-year duration, 18% of all trials were terminated. Comparing the early/later 5-year periods, 21% of initiated trials versus 15% of the trials respectively, were terminated – an indication that the industry is improving in its overall clinical trial success rate.

Across all 10 years, the top two reasons for terminating a trial (by absolute trial counts) were lack of efficacy and poor enrollment. On a relative basis (as percentage of total terminated trials), 27% of the trials were terminated due to missing efficacy endpoints, and 26% due to slow/inadequate enrollment. The third most frequently cited reason for terminating the trials was a higher than expected incidence of serious adverse events – just 10% of all trials ended for this reason.

(See Table 1 at end).

Signs of Industry Lessons Learned

A comparison of the relative proportion of terminated trials between the early/later 5 year durations reveals several important trends (Figure 1a, b).

Efficacy:  The first half of the analysis period saw approximately 30% of Phase II, II/III and III trials terminated for missing an efficacy endpoint. During the second half of the decade, terminated Phase II trials for this reason dropped by approximately 5%, and Phase III trials dropped from 29.6% to 13.3%. This significant decrease may be attributed to better go/no-go decisions being made earlier, better trial design, or to an increase in trials involving targeted therapies.  When comparing the percentage of Phase II/III trials terminated due to lack of efficacy, the trend is reversed, with a 10% overall increase in the more recent five year period. But due to the low overall numbers of trials, 9 vs. 6, we hesitate to over-interpret this observation.

Poor or slow enrollment:  For both Phase II and III terminated trials, an overall decrease in terminations attributed to poor/slow patient enrollment is evident for the more recent five year time period.  One reason for this trend might be attributed to sponsor’s allowing enrollment periods to lengthen in order to reach a desired level of patient accrual. Using trial-level enrollment period data published in Trialpredict, we compared average enrollment periods for completed, industry-sponsored breast cancer trials (the disease area with the greatest numbers of terminated trials) and found that an increase in overall enrollment period duration (combined for Phase II, II/III and III) is evident:  22.8 months versus 29.2 months for the early and later five-year comparison periods, respectively.

A decrease in attrition due to poor enrollment may also reflect improvement in site and investigator selection.

Toxicity or serious adverse events:  In terms of absolute numbers of trials stopped early for this reason, it would appear that this improved during the most recent five year period (see Table 1). But when examined as a proportion of trials by phase (Figure 2), the overall trend toward improvement is diluted.

Strategic Business Decision:  It is very often the case that a company will proactively terminate a trial based on negative results obtained from full analysis of trials that read out after the start of the current trial, or  due to safety issues seen in another concurrent or recently completed trial, or based on stronger efficacy signals in another patient population or disease. These decisions have little to do with an individual trial, but rather reflect company strategy to better align a drug program for success. The earlier the sponsor makes the no-go decision, the earlier unnecessary spend can be eliminated.

When all trials terminated due to a business decision (drug strategy shift, pipeline reprioritization) are combined, there is a significant positive trend to terminate for strategic reasons (Figure 3). This is a clear signal that the industry is taking a hard look at their development programs.  This is particularly true for the larger biopharma companies. Supporting this conclusion is the uptick in percent of terminated trials reportedly terminated due to funding issues (Figure 4). For Phase II trials, a 10-fold increase was seen between the two time periods, and a three-fold increase for Phase III trials.

Hint of Trouble in Paradise?

Amidst all of these reasons to celebrate is one troubling observation:  a sharp decline in trials terminated due to early positive outcomes (Figure 5).  Absolute trial counts are relatively low (23 total trials in ’02 – ’06 compared to 13 for the more recent period), but on a relative basis the drop from 11.1% to 6.7% of trial per phase is significant. 

Lessons-Learned ‘Who’s Who’

The companies that sponsored 10 or more terminated trials over the past decade, and the breakdown, by overall percentage of total trials for the two half-decade periods reveals a not-so surprising list of players [Figure 6]. For simplicity of comparison, the data sums trials sponsored by companies acquired by the major players during the analysis periods, and omits trials jointly sponsored by two or more companies.  So amongst these sponsors, who show improvement over time? Eli Lilly, Sanofi, J&J, Amgen and Takeda all demonstrated a decrease in relative percentages of trials terminated. AstraZeneca had a smaller relative decrease overall compared to the aforementioned companies.

Honing in on the major company sponsors, the trends in reasons for termination do not differ from those reported for all industry-sponsored trials; the top reasons include slow/poor enrollment, lack of efficacy, incidence of severe adverse events, and strategic business decisions (Table 2).

Lack of Efficacy:
For the ’02 – ’06 time period, Pfizer (18 trials/2.0%) led the pack in relative percentage of trials terminated due to lack of efficacy, followed by Novartis (12 trials/2.0%), GSK (9 trials/1.5%) and BMS (8 trials/1.3%). For the ’07 – ’11 five year period, Novartis moved to first place (14 trials/ 3.0%), bumping Pfizer to second (13 trials/2.8%).

Poor Enrollment:
The sponsors with a significant downward trend in trials terminated due to slow or poor enrollment include Eli Lilly (4.4% to 0.4%), Sanofi (3.1% to 1.7%) and Novartis (2.9% to 1.7%). Those with increases in relative terminations for this reason include BMS, Bayer AG, and GSK.

Safety/Adverse Events:
No downward trends in terms of overall company performance can be gleaned from the dataset. Both Pfizer and GSK have upticks in relative percentages for trials terminated due to toxicity or adverse events.

Strategic Business Decisions:
Six of the thirteen major pharma companies demonstrated a downward trend in the relative percentage of trials terminated due to strategic decision-making (Figure 7).  Of more interest are those companies that slipped in this category: Amgen and GSK.  Of note is the observation that neither Bayer AG, Merck & Co. nor Eli Lilly reported any trials terminated across the entire ten year analysis period, due to proactive business decisions. Perhaps these sponsor companies already ‘honed’ their internal decision-making processes?

Early Positive Outcome: For this best of all reasons to terminate a clinical trial, despite the overall negative trend reported across all sponsors, several of the pharma sponsors show recent positive trends (Figure 8). Comparing absolute trial counts between the two five-year periods, Roche shows a decrease from 3 trials to 1, while Celgene reports a decrease from 2 trials to 1. Novartis, Sanofi, J&J and Merck & Co, each had one trial terminated in both periods.  Bayer AG had no reported terminations for early positive outcomes in the early part of the decade, but during the second half reported 2 such trial terminations.  Why the drop?  Perhaps trial design complications preclude the possibility of early positive terminations?  Alternatively, might the move to targeted therapies demand full trial read out?

Conclusions:
The numbers of industry-sponsored trial terminations reported during 2007 – 2011 decreased in comparison to those reported during 2002 – 2006. In addition to this overall positive trend, it appears that company sponsors are proactively taking steps to terminate more trials, or programs in recent years. Taking such steps can only help to improve the balance sheets, saving more R&D dollars for other more promising trials and programs.

Enrollment issues continue to plague the industry, but the five year period comparison does reveal signs of progress. With the emergence and uptake of electronic trial management software solutions, it may be that sponsors can recognize poor performing sites or investigators sooner and replace them earlier. Alternatively, it may be that more realistic timelines for trials are on the rise.

Futility, or lack of efficacy also is a persistent issue in clinical trial performance, but a strong sign of improvement in this area is the significant reduction in Phase III trials terminated for this reason during the second half of the analysis period.  This suggests that better decisions are being made during or after Phase II completion resulting in a higher Phase III success rate. Due to the length of Phase III Oncology trials, Citeline will need to revisit this analysis after several more years of data have accumulated.

Table 1.Terminated Oncology Trial Counts by Five Year Period and Trial Phase (data accessed 06/20/2012)

1. The data from Trialtrove originates from an export taken on June 20, 2012. Over time, it is anticipated that additional trial start dates will be disclosed in the public domain.