METHODS

RESULTS

We originally planned a study size of 300 and assumed that 20% of subjects would not provide analysable end point data, and therefore asked six centres to enrol 50 patients each. However, one centre never commenced enrolment, and a higher non-analysability rate of 30% was observed in the first 150 patients. We therefore expanded the study size to 350 and added a further six centres, each to recruit 25 patients. A total of 354 patients were entered according to study criteria and received treatment; 120 patients were randomised to podophyllotoxin solution, 118 to podophyllotoxin cream, and 116 to podophyllin. The study population comprised 182 (51%) men, 95% were white, 56% were smokers, 69% had less than 10 genital warts, and 96% of males were uncircumcised. The characteristics of the subjects in the three treatment arms were well matched at entry (table 1⇓), except that there were slightly more circumcised men in the podophyllotoxin solution arm (9, compared to podophyllin 3, podophyllotoxin cream 1, p=0.02). 276 (78%) of patients completed the trial according to the protocol, the principal analysis populations being 96, 82, and 98 subjects in each of the podophyllotoxin solution, podophyllotoxin cream and podophyllin arms respectively. The only significant difference between the principal analysis population and defaulters was that smokers were over-represented in the latter group (52% v 70% respectively, p=0.004).

Tables 2 and 3⇓⇓ present both the raw complete remission rates and odds of achieving remission, by treatment group and population (the principal analysis, worst case, and best case populations), for original warts and all warts respectively. Treatment was associated with the odds of remission, with increased odds for both podophyllotoxin solution (for original warts OR 2.88, 95% CI 1.52 to 5.45; for all warts OR 2.93, 95% CI 1.56 to 5.50) and for podophyllotoxin cream (for original warts OR 1.67, 95% CI 0.89 to 3.15; for all warts OR 1.97, 95% CI 1.04 to 3.70) compared to podophyllin. In the analysis where subjects defaulting from follow up were assumed to have been cured we again observed increased odds both for podophyllotoxin solution (for original warts OR 2.91, 95% CI 1.59 to 5.34; for all warts OR 3.04, 95% CI 1.68 to 5.49) and for podophyllotoxin cream (for original warts OR 2.14, 95% CI 1.19 to 3.82; for all warts OR 2.46, 95% CI 1.38 to 4.40) compared to podophyllin. These differences therefore reached statistical significance in all analyses for podophyllotoxin solution, but only reached statistical significance for podophyllotoxin cream in the best case analysis where defaulters were assumed to be cured, and in the principal analysis for all warts. In all analyses, males and those with fewer than 10 warts at baseline had significantly increased odds of remitting. Smoking was not associated with the odds of remission, except when considering all warts in the best case analysis when smokers had significantly increased odds of remission. As smokers were significantly over-represented among subjects who defaulted we do not assume this to be of relevance.

We measured the relative reduction in the number of original warts and all warts over the 4 week treatment period for the three treatment groups. For original warts podophyllotoxin solution (90.0%, 95% CI 85.3 to 94.6) was significantly superior to podophyllin (76.2%, 95% CI 67.4 to 85.1) and podophyllotoxin cream (75.4%, 95% CI 66.5 to 84.3), whereas for all warts podophyllotoxin solution (83.7%, 95% CI 77.0 to 90.4) was only significantly better than podophyllin (62.2%, 95% CI 46.6 to 77.7) (podophyllotoxin cream 64.1%, 95% CI 48.1 to 80.2).

Local side effects were seen in 33%, 24%, and 17% of the podophyllotoxin solution, cream, and podophyllin groups and ulceration in 18%, 12%, and 10%, respectively. Forty five per cent of subjects who were clear of warts at 4 weeks returned for a 12 week follow up or earlier in case of relapse. Relapse rates of 15/33 (45%) in the podophyllotoxin solution group, 12/22 (55%) in the podophyllotoxin cream group, and 5/19 (26%) in the podophyllin group were observed. These differences were not statistically significant. We suggest these figures should be regarded as merely descriptive because of (a) the high proportion of defaulters in each group, and (b) the selection bias for re-attendance if further warts occurred.

DISCUSSION

This is one of the largest randomised trials of treatments for genital warts yet reported. Its size, design, and multicentre nature make it likely that the results are a reasonable reflection of treatment outcomes for first episode genital warts in clinical practice. The patient population was mainly heterosexual, 96% of males were uncircumcised as is usual in the United Kingdom, and the majority of lesions in these men were subpreputial or on the distal penis. The three treatment arms were well matched at entry with no significant differences of relevance.

Although podophyllotoxin solution and gel are generally available at 0.5% strength, evaluations of podophyllotoxin cream have reported the use of 0.5%, 0.3%, and 0.15% strengths.^11,12,17 Podophyllotoxin 0.3% cream appeared to be associated with a higher frequency of adverse events than 0.15% cream, and it was this latter licensed preparation that we studied.¹² Our data can be compared with other published comparative trials of podophyllin and podophyllotoxin solution and cream.^11,12,14–,¹⁷ We found generally similar clearance rates and differences in response rates between the different treatments to the published literature, although the numbers of subjects we studied was larger than those previously reported.

Patients with genital warts in clinical trials not infrequently default from follow up, which previous reports have observed.^12,14 We explored ways of utilising the data set in a manner where differing assumptions could be made about these “missing values.” Sexually transmitted infections in general, and genital warts in particular, have been shown to be associated with psychological morbidity and concepts of shame.¹⁹ The clinicians who conducted the study were therefore of the opinion that defaulters were more likely to be cured than have persisting disease. The best case analysis therefore has substantial justification, and approaches the real life situation. We also present the response rates for the podophyllotoxin preparations in a novel manner as odds ratios by multivariate regression compared to the efficacy of podophyllin. This illustrates these comparisons (for example, for all warts in the best case analysis) in quite a striking manner, with podophyllotoxin solution and cream being 3.0 and 2.5 times more likely to produce cure than podophyllin. Podophyllotoxin solution was usually associated with numerically better cure rates than podophyllotoxin cream although in none of the regression analyses did this reach significance. Considering that we were comparing a 0.5% podophyllotoxin formulation with a 0.15% formulation the degree of similarity in treatment effects is of note. Although podophyllotoxin solution emerged as significantly superior to podophyllin in all regression analyses, podophyllotoxin cream was only statistically significantly better than podophyllin in the principal analysis for all warts and the best case analysis. Even larger trials than ours would be needed to settle this latter issue definitively. The data with regard to treatment side effects are suggestive of a correlation between side effects and efficacy, as has been observed previously, although the rates of side effects we observed were not statistically different between treatments.¹⁶ Side effects were not a limiting factor for treatments within the trial. Recurrence rates after individual treatments are crucial to overall clinical efficacy in real life. Our ability to reach conclusions about differences in recurrence rates was limited by the rate of patient default we observed at the 12 week visit, and we do not ascribe any significance to these data.

The economic evaluation was also complicated by the frequency of discontinuations during treatment and the reduced proportion of previously cleared patients who returned for the last follow up visit. Health economic analyses sometimes use disease free or healthy days as an effectiveness measure.²⁰ In such studies of genital wart therapy it is not possible to obtain data for disease free days in all patients, and we therefore chose to use complete remission of all warts as the most appropriate effectiveness measure. Arguments can be made concerning the relevance of including indirect costs in such evaluations.^21,22 In this study the final results regarding which treatments were more cost effective do not differ whether indirect costs are included or not.

In economic evaluations both average costs, effects, and cost effectiveness should be demonstrated, as well as the marginal rates—that is, the additional cost per additional unit of the effect expressed as the marginal cost effectiveness rate. In the main economic analysis the total costs were lower for one of the evaluated drugs (podophyllotoxin solution) compared with the comparator drug (podophyllin solution) while its treatment effects were superior. It is therefore not necessary nor meaningful to estimate the marginal cost effectiveness rates for podophyllotoxin solution because it is clear that this alternative is more favourable. For podophyllotoxin cream the marginal cost in the main analysis was higher (£37.75) (table 4⇑), but with a better clinical effect expressed as complete remission rate (62.2% compared with 46.9%) (table 5⇑) resulting in a marginal cost effectiveness rate of £246.73 per further cured patient. Podophyllotoxin cream is therefore a cost effective alternative compared with podophyllin solution since the cost of curing one further patient with podophyllotoxin cream is considerably lower than the average cost of curing one patient with podophyllin. In these analyses we have also used the confidence intervals for the complete remission rates associated with the three treatment alternatives as a form of sensitivity analysis. Although economic analyses are not usually evaluated with formal tests of statistical significance, there appears to be a hierarchy of total costs per complete remission with podophyllotoxin solution as the most cost effective and podophyllin as the least cost effective. It is believed that clinicians and patients arrive at choices between podophyllotoxin solution and cream on the grounds that podophyllotoxin cream is a more acceptable treatment for patients with “non-penile anogenital warts.” Although it seems common sense that the superior ease of application of cream in less accessible anogenital sites would result in superior acceptability and perhaps effectiveness this has not been formally demonstrated. In the light of this assumption we suggest that the strategy of using podophyllotoxin solution for penile warts and podophyllotoxin cream for other anogenital warts is a cost effective one.

A few previous economic analyses of genital wart therapy have been published.^23–,²⁵ As in our work, these studies illustrated that one of the central determinants of overall costs is the frequency of clinic visits, and this is inherently reduced with patient applied therapy. An economic evaluation should preferably reflect real clinical practice and it is therefore necessary to identify resource use for the purposes of the trial itself.²⁶ In our trial the patients using podophyllotoxin were asked to attend every week. Assuming only two visits, one at the start of treatment and one after 4 weeks, as in normal clinical practice, would reduce total average costs for patients treated with podophyllotoxin solution and cream by 24% and 22%, respectively.

Although self treatment regimens were developed and analysed some years ago,¹⁰ it is only recently that such therapies have become widely available. Podophyllotoxin is now marketed as a solution in many countries worldwide, and also as a cream formulation in Europe. In North America podophyllotoxin is also available in a 0.5% gel formulation, but the only published evaluation of this preparation in adults with genital warts suggested lower clearance rates that we observed for podophyllotoxin solution and cream.¹³ The primary clearance rates for podophyllotoxin observed in our study of ∼70% in both sexes compare well with other treatments.⁴ Both US and UK national guidelines for the treatment of genital warts continue to recommend the use of podophyllin.^27,28 Our demonstration that treatment of anogenital warts with podophyllotoxin solution and cream shows greater efficacy and cost effectiveness than clinic based treatment with podophyllin, as well the recognised pharmacological deficiencies of podophyllin, suggest that these guidelines may need further modification in due course.^8,9 However, in clinical practice one of the foremost problems remains recurrence after therapy. Our study observed substantial recurrence rates similar to previous studies of podophyllin and podophyllotoxin, but also illustrates how recurrence after therapy diminishes cost effectiveness. Therefore, treatments, either single or combination, are needed which combine high rates of primary clearance with low recurrence rates. Recently interest has focused on immunological mechanisms of regression and it is possible that therapies associated with lower recurrence rates may be invoking such cytotoxic immune responses.²⁹ If new treatments satisfying these clinical imperatives are developed they should be evaluated to address the issues of cost effectiveness that we have outlined.