Ideally, the promotion and use of any drug should be based on evidence of efficacy and safety. Such evidence is not always easily obtained. This is particularly the case for many common chronic diseases where the ultimate outcomes, such as cure, permanent disability or death, require lengthy longitudinal studies. In these situations, intermediate outcome measures which are thought to predict clinically significant endpoints are used. Examples of such intermediate outcomes include lowering cholesterol in the hope that this will decrease myocardial infarctions, or measuring bronchodilatation with the expectation that improvement will be associated with normalisation of airway function in the long term. When deciding whether new drugs should take a significant role in their practice, prescribers need to ask 3 questions about these intermediate outcomes:
- are they clinically relevant?
- are they valid predictors of final outcomes?
- is any statistically significant difference also clinically significant?
This issue of Australian Prescriber reviews a number of drugs which may produce a modest benefit for patients with benign prostatic hypertrophy. Two of these drugs, prazosin and finasteride, are generations apart, both in development and mode of action. Prazosin is a relatively cheap drug that is no longer under patent, while finasteride has been developed specifically for the treatment of benign prostatic hypertrophy. The efficacy of these drugs has been assessed by an analysis of intermediate outcomes. This analysis illustrates some of the problems with surrogate endpoints.
For benign prostatic hypertrophy, an assessment of outcome should include a reduction in the need for surgery. However, finasteride has only been shown to affect intermediate outcomes such as urinary flow, prostatic volume and symptom score. These outcomes are easily measured, but small differences have only been revealed by large studies. This implies that the effects of the drug are small.
What can the prescriber deduce from such information?
Are the intermediate outcomes valid predictors of a clinically important endpoint? Prostatic volume does not correlate well with the degree of urinary obstruction. On the other hand, urinary flow rates, intuitively, seem more important. However, is a small change in urine flow also associated with a reduction in symptoms such as frequency and nocturia? These outcomes are far more relevant to patients.
Is the change in the measured outcome clinically significant? Finasteride produced a small but statistically significant change in symptom scores. Such a change may go unnoticed by the individual patient and may not result in an improvement in important but unscored symptoms such as nocturia. Finasteride also produces a statistically significant increase in peak urinary flow of up to 17% (9.6 to 11.2 mL/second). In contrast, surgery produces an increase of 100%. If a patient passed 1500 mL of urine a day at peak urinary flow rate, he may be able to gain an extra 23 seconds away from the toilet. Finasteride costs approximately $2.00 per day, so the patient could expect to accrue an extra 24 hours away from his toilet after 10 years' treatment, at a cost of $7500.
There is little doubt that finasteride can improve some outcome measures; however, the benefits are, at best, modest. It is not surprising that there have been no published comparative trials with its cheaper competitor, prazosin. Such direct comparisons would enable prescribers to make decisions based on both efficacy and cost.
The challenge for pharmaceutical companies is to produce and supply data based on meaningful outcomes using current therapies that are valid comparators. As prescribers, we are aware that it is the patient, not the statistic, that stands in front of the urinal.
Further reading
Lange PH. Is the prostate pill finally here? [editorial; comment]. N Engl J Med 1992;327:1234-6.
Gormley GJ, Stoner E, Bruskewitz RC, ImperatoMcGinley J, Walsh PC, McConnell JD, et al. The effect of finasteride in men with benign prostatic hyperplasia. The Finasteride Study Group [see comments]. N Engl J Med 1992;327:1185-91. Comment in: N Engl J Med 1992;327:1234-6. Comments in: N Engl J Med 1993;328:442-3,443.
