Keys 2003.

Study characteristics
Methods	Multicentre RCT conducted in the US (GOG #71, RTOG #84‐12). Primary study objective was to determine whether adjuvant hysterectomy following radiotherapy for women with bulky Stage IB cervical cancer improved survival.
Participants	256 eligible women. It was not mentioned whether the women were evaluated clinically or radiologically after radiotherapy in order to assess the tumour response and residual disease. Surgical staging of lymph nodes was optional and was performed on 57 (22%) women, equally divided between the study arms. Any women with metastasis to the para‐aortic nodes were ineligible for the RCT. Histopathology type Adenocarcinoma: 6% in the radiotherapy group and 7% in the radiotherapy + hysterectomy group. Adenosquamous carcinoma: 7% in the radiotherapy group and 7% in the radiotherapy + hysterectomy group. Squamous cell carcinoma: 86% in the radiotherapy group and 86% in the radiotherapy + hysterectomy group. Age ≤ 30 years: 9% in the radiotherapy group and 9% in the radiotherapy + hysterectomy group. 31–40 years: 37% in the radiotherapy group and 29% in the radiotherapy + hysterectomy group. 41–50 years: 32% in the radiotherapy group and 35% in the radiotherapy + hysterectomy group. 51–60 years: 17% in the radiotherapy group and 18% in the radiotherapy + hysterectomy group. 61–70 years: 4% in the radiotherapy group and 8% in the radiotherapy + hysterectomy group. 71–80 years: 1% in the radiotherapy group and 2% in the radiotherapy + hysterectomy group. Race White: 49% in the radiotherapy group and 52% in the radiotherapy + hysterectomy group. African American: 35% in the radiotherapy group and 27% in the radiotherapy + hysterectomy group. Other: 15% in the radiotherapy group and 21% in the radiotherapy + hysterectomy group. GOG performance grade 0: 77% in the radiotherapy group and 76% in the radiotherapy + hysterectomy group. 1: 22% in the radiotherapy group and 20% in the radiotherapy + hysterectomy group. 2: 1% in the radiotherapy group and 4% in the radiotherapy + hysterectomy group. Tumour size 4 cm: 9% in the radiotherapy group and 13% in the radiotherapy + hysterectomy group. 5 cm: 36% in the radiotherapy group and 33% in the radiotherapy + hysterectomy group. 6 cm: 32% in the radiotherapy group and 27% in the radiotherapy + hysterectomy group. 7 cm: 10% in the radiotherapy group and 18% in the radiotherapy + hysterectomy group. ≥ 8 cm: 12% in the radiotherapy group and 9% in the radiotherapy + hysterectomy group. Tumour type Exophytic: 48% in the radiotherapy group and 43% in the radiotherapy + hysterectomy group. Barrel: 52% in the radiotherapy group and 57% in the radiotherapy + hysterectomy group.
Interventions	Radiotherapy alone (external and brachytherapy) Radiotherapy + hysterectomy (attenuated irradiation + extrafascial hysterectomy) Radiotherapy Radiotherapy was delivered externally. Daily fraction size was 180 Gy and external treatment carried to a total dose of 40 Gy for the radiation alone regimen and 45 Gy for the adjuvant hysterectomy regimens. Dose distribution could not vary > 5% across the treatment volume. These parameters were the same for both treatment arms. The intracavitary treatment dose prescription was different between the treatment arms. Both groups were to have an intracavitary treatment 1–2 weeks after completing external treatment. The radiotherapy alone group received 40 Gy to 'point A', while those who were to have hysterectomy received only 30 Gy. Interstitial therapy was not permitted on this protocol. A total dose of 80 Gy to 'point A' was prescribed for the radiotherapy alone regimen and 75 Gy for the adjuvant hysterectomy regimen. A minimum dose of 55 Gy was prescribed to 'point B' for both regimens. A parametrial boost was permitted if necessary to achieve this dose. All irradiation was to be completed within 10 weeks. Hysterectomy The radiotherapy + hysterectomy group was then to undergo extrafascial hysterectomy with removal of tubes and ovaries, if present, 2–6 weeks after completion of all irradiation. This operation was described to include the removal of the corpus and cervix without contiguous parametrial tissue.
Outcomes	OS Pelvic‐free survival Pelvic recurrence
Notes	256 eligible women with exophytic or 'barrel' shaped tumours measuring 4 cm were randomised to either external and intracavitary radiotherapy (124 women) or attenuated radiotherapy + extrafascial hysterectomy (132 women).   Of the women randomised to receive radiotherapy alone, 87% received a total 'point A' dose of 78 Gy or more while 49% had a total duration of treatment of ≤ 60 days. For the radiotherapy + hysterectomy treatment regimen, 90% received a total 'point A' dose of ≥ 71 Gy while 80% had a total duration of treatment of ≤ 60 days. Regarding the proportion of received to prescribed dose to 'point A', 96% of women had proportions > 0.80 from the radiotherapy alone group and 95% of women from the radiotherapy + hysterectomy regimen. 22% of women (28/124 (23%) women in radiotherapy alone group and 29/132 (22%) women in radiotherapy + hysterectomy group) received the optional up‐front surgical staging. Of the 103 women who did not have prerandomisation surgical staging on the hysterectomy regimen, 54 (52%) had a hysterectomy and lymph node sampling procedure. Of these, 7 (13%) had positive para‐aortic nodes. Both treatment programmes were well tolerated. Hysterectomy did not increase the frequency of reported grade 3 and 4 adverse effects (13 (10%) women for each regimen). 18/26 of these serious adverse effects were from the gastrointestinal or genitourinary tract exclusively. The frequency of any reported adverse effect was higher for the radiotherapy + hysterectomy group (56% with radiotherapy alone vs 63% with radiotherapy + hysterectomy). 57 (46%) women had progression of disease among the radiotherapy alone group and 49 (37%) women had progressions in the radiotherapy + hysterectomy group. The PRS results included, as failures, those with disease progression and deaths that occurred without progression (8 with radiotherapy alone vs 11 with radiotherapy + hysterectomy). Significance level for the log‐rank test was P = 0.07 (1‐tail). Reduction in the risk of progression/death for the radiotherapy + hysterectomy group to the radiotherapy alone group was 23% (i.e. RR 0.77, 90% CI –3 to 43). Median PFS was 7.4 years for the radiotherapy alone regimen and no estimate was available for the radiotherapy + hysterectomy regimen (PFS was 53% at 8.4 years). There were 2 treatment‐related deaths in the radiotherapy alone group and 1 in the radiotherapy + hysterectomy group. 8 women died of intercurrent disease and 12 died of unknown causes. Those who were last seen alive had a median follow‐up time of 9.6 years (range 0.3 to 16.1 years). 12 women in each regimen (10% in radiotherapy alone group vs 9% in radiotherapy + hysterectomy group) were lost to follow‐up by 5 years. There was no apparent difference in the survival experience of women by treatment regimen (P = 0.26, 1‐tail). The relative risk estimate of the combined‐treatment group to the radiotherapy alone group was 0.89 (90% CI 0.65 to 1.21). 56 women died in each treatment regimen. The modelling of survival identified the same prognostic factors with very similar relative risk estimates. Adjusting for tumour size, performance status and age, the relative risk of progression for the combined treatment group to the radiotherapy alone group was statistically significant (i.e. RR 0.72; P = 0.04, 1‐tail). The adjusted relative risk estimate of death was 0.84, which was not statistically significant. At 5 years, the strictly local relapse incidence was up to 27% for the radiotherapy alone regimen vs 14% for the radiotherapy + hysterectomy regimen, although the radiotherapy + hysterectomy group was slightly more likely to have distal progression (16% with radiotherapy alone vs 20% with radiotherapy + hysterectomy). These rates agreed with the distribution of progression site. To determine if there was any differential effect of adjuvant hysterectomy by tumour size, a test for interaction was performed. The interaction term, when quantified using the whole centimetre tumour size, was of borderline significant (P = 0.06) for PFS but was statistically significant (P = 0.007) for survival. The results indicated that women in the radiotherapy + hysterectomy group had a lower risk of progression and death than the radiotherapy‐alone group for tumour sizes of 4 cm, 5 cm and 6 cm. Combined, the RR for death was 0.60 (relative risk of progression 0.58) for the radiotherapy + hysterectomy group to the radiotherapy alone group. Women with tumour sizes of ≥ 7 cm in the radiotherapy + hysterectomy group progressed at a higher rate after 14 months; the relative risk estimate was 1.27 (relative risk of death 2.03). The crossover from lower to higher risk for the radiotherapy + hysterectomy regimen occurred between 6 cm and 7 cm for both PFS and survival. Histological evaluation of the cervical specimens for the 123 women who underwent hysterectomy identified 59 (48%) with no detectable malignancy, while 49 (40%) were microscopically positive and 15 (12%) were grossly positive. Median time to surgery from study entry was 3.0 months (10th percentile 2.4 and 90th percentile 3.8 months). Comparing PFS and survival of women with negative vs positive specimens were statistically significant. Women with grossly positive hysterectomy specimens progressed and died at almost 7 times the rate compared to those with negative specimens.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Randomization with equal probability of assignment to each treatment regimen was carried out by a block arrangement balancing the treatment assignment within major GOG institutions and the option of para‐aortic lymph node sampling".
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not possible to blind participants and clinicians to these interventions.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not reported.
Incomplete outcome data (attrition bias) All outcomes	Low risk	256/256 eligible women were analysed for primary survival outcomes using appropriate statistical techniques. 2 women in each arm did not receive radiotherapy and were not examined for adverse effects.
Selective reporting (reporting bias)	Low risk	All pertinent outcomes appeared to have been reported by the trial authors.
Other bias	Unclear risk	Insufficient information to assess whether an additional risk of bias existed.