Forty-two studies were identified (n at least 57,870): three RCTs (n=2,639, range 478 to 1,477) and 39 non-randomised studies (n at least 55,231). The non-randomised studies included 24 treatment studies (n at least 6,401, range 103 to 721) and 15 long-term toxicity studies (n at least 48,830, range 230 to 40,576). All three RCTs had a non-inferiority design, 90% power and unclear details of loss to follow-up. Two RCTs performed intention-to-treat and per protocol analysis. One RCT described the method of randomisation clearly.
Survival: All three RCTs found no significant differences between treatment groups in relapse-free survival after five years.
Non-randomised studies found that five-year overall survival did not differ greatly for different management strategies: surveillance, range 97 to 100% (four studies); radiation therapy, range 95 to 100% (six studies); and carboplatin, range 94 to 100% (five studies). Relapse patterns varied for the different strategies. Most patients under surveillance and treated with adjuvant carboplatin had para-aortic nodal relapse; 2% to 3% of patients treated with para-aortic radiotherapy were noted to have pelvic nodal relapse.
Toxicity: Results were reported for acute toxicity in two RCTs. Results for long-term toxicity were reported for the non-randomised studies including second malignancy (12 studies), cardiac toxicity (two studies) and quality of life (one study).