Stockholm III
Needs reviewA prospective randomized trial on different preoperative radiotherapy regimens in rectal cancer, Stockholm III
Phase 3N = 840NCT00904813
- Population
- Patients with biopsy-proven resectable rectal adenocarcinoma planned for abdominal resection
- Design
- Multicentre, randomised, non-blinded, phase 3, non-inferiority trial in patients with biopsy-proven rectal adenocarcinoma planned for abdominal resection, from 18 Swedish hospitals. Patients were randomised to 5 x 5 Gy with surgery within 1 week (short-course radiotherapy, SRT), 5 x 5 Gy with surgery after 4-8 weeks (SRT with delay), or 25 x 2 Gy with surgery after 4-8 weeks (long-course radiotherapy with delay, LRT-delay). After a protocol amendment, hospitals could randomise to all three treatments or to the two short-course treatments only. 840 patients were randomised between October 1998 and January 2013: 385 in the three-arm randomisation (129, 128 and 128) and 455 in the two-arm randomisation (228 and 227). Intention-to-treat analysis.
- Primary endpoint
- Local recurrence
- Arms
- Short-course radiotherapy, immediate surgery (control, n = 357): 5 x 5 Gy with surgery within 1 week
- Short-course radiotherapy, delayed surgery (experimental, n = 355): 5 x 5 Gy with surgery after 4-8 weeks
- Long-course radiotherapy, delayed surgery (experimental, n = 128): 25 x 2 Gy with surgery after 4-8 weeks
- People
- Erlandsson J
Local recurrenceHR 1.44Local recurrenceHR 2.24Δ Surgical complications−12 pointsΔ median OS+2.2 yearsLocal recurrenceHR 1.31Local recurrenceHR 2.23
Show the numbers
| Endpoint | Analysis | Short-course radiotherapy, delayed surgery / Long-course radiotherapy, delayed surgery | Short-course radiotherapy, immediate surgery | Effect |
|---|---|---|---|---|
| Local recurrenceIntention-to-treat analysis (840 randomised)Time to local recurrence (primary endpoint): short-course radiotherapy with delay versus short-course radiotherapy | primaryErlandsson 2017 | — | — | HR 1.44Hazard ratio versus short-course radiotherapy (SRT-delay is the experimental arm in the 2021 report). The 2017 abstract prints the interval as a 95% CI of 0.41-5.11, but it also defines the non-inferiority criterion on a double-sided 90% CI, and the open-access 2021 report states that its local recurrence intervals are 90% CIs; the 2017 full text could not be opened, so the level of the printed interval is unconfirmed. Cumulative incidence of local recurrence in the whole trial: 8 of 357 patients after short-course radiotherapy, 10 of 355 after short-course radiotherapy with delay and 7 of 128 after long-course radiotherapy with delay; p=0.48 is printed once for both comparisons. The abstract states that both delayed regimens were deemed non-inferior, although the upper limits of both printed intervals exceed the 1.7 margin it states; the 2021 report explains that the margin was re-set to 6.5 after the trial began. |
| Local recurrenceIntention-to-treat analysis (840 randomised)Time to local recurrence (primary endpoint): long-course radiotherapy with delay versus short-course radiotherapy | primaryErlandsson 2017 | — | — | HR 2.24Hazard ratio versus short-course radiotherapy; the 2017 abstract prints the interval as 0.71-7.10 without naming the level (see the note on the other comparison). |
| Surgical complicationsThree-arm randomisation (129, 128 and 128 patients)Postoperative complications (three-arm randomisation) | primaryErlandsson 2017 | 38% | 50% | P=0.075Odds ratio versus short-course radiotherapy: 0.59 (95% CI 0.36-0.97) for short-course radiotherapy with delay and 0.63 (0.38-1.04) for long-course radiotherapy with delay. In the pooled comparison of the two short-course regimens the risk of postoperative complications was lower after short-course radiotherapy with delay (odds ratio 0.61, 95% CI 0.45-0.83, p=0.001). The abstract prints the counts as '144 [53%] of 355 vs 188 [41%] of 357'; the percentages do not match the counts (144/355 = 40.6%, 188/357 = 52.7%), and the odds ratio calculated from the counts (0.61) agrees with the printed one, so the percentages appear transposed; only the odds ratio is relied on. Acute radiation-induced toxicity was recorded in 1 of 357 patients (<1%) after short-course radiotherapy, 23 (7%) of 355 after short-course radiotherapy with delay and 6 (5%) of 128 after long-course radiotherapy with delay. |
| OSStockholm III randomised patients (357, 355 and 128 in the whole trial); the abstract does not say which analysis the three medians come from (the three-arm randomisation, 129, 128 and 128 patients, is the only one that includes all three groups)Median overall survival after short-course radiotherapy, short-course radiotherapy with delay and long-course radiotherapy with delay; the abstract gives 'range' values, not confidence intervals | long term-updateErlandsson 2021 | median 10.3 years | median 8.1 years | In the short-course comparison the median overall survival was 8.1 years after short-course radiotherapy and 10.2 years after short-course radiotherapy with delay. The median follow-up for overall survival is 9.8 years (IQR 7.7-12.6) in the full text; the 5.7 years in the abstract is its overall follow-up figure. In the three-arm randomisation local recurrence occurred in 3 of 129, 4 of 128 and 7 of 128 patients and distant metastases in 31, 38 and 38 patients; in the short-course comparison local recurrence occurred in 11 of 357 and 13 of 355 patients and distant metastases in 88 and 82. No comparison showed a statistically significant difference. Full-text hazard ratios for overall survival with short-course radiotherapy as reference: short-course comparison 0.84 (95% CI 0.66-1.06, P=0.146); three-arm randomisation 0.75 (0.51-1.09, P=0.137) for short-course radiotherapy with delay and 0.88 (0.61-1.29, P=0.512) for long-course radiotherapy with delay. |
| Local recurrenceShort-course radiotherapy comparison (357 and 355 patients), analysed by intention to treatTime to local recurrence: short-course radiotherapy with delay versus short-course radiotherapy (reference) | long term-updatemedian follow-up 5.7 yearsErlandsson 2021 | — | — | HR 1.31P=0.518Interval 0.66-2.61; the Methods state that local recurrence intervals are 90% confidence intervals. The Methods also state that the trial was designed as non-inferiority with a margin of 1.7 for the upper limit of a one-sided 90% confidence interval, and that after the trial started a new power calculation was done because local recurrence was less frequent than expected, so that non-inferiority could be decided at an upper limit of 6.5. In the three-arm randomisation the hazard ratio for short-course radiotherapy with delay was 1.18 (0.33-4.15, P=0.830). |
| Local recurrenceThree-arm randomisation (129, 128 and 128 patients), analysed by intention to treatTime to local recurrence: long-course radiotherapy with delay versus short-course radiotherapy (reference) | long term-updatemedian follow-up 5.7 yearsErlandsson 2021 | — | — | HR 2.23P=0.236Interval 0.70-7.14 (90% confidence interval for local recurrence, per the Methods). |
| Endpoint | Analysis | Short-course radiotherapy, delayed surgery / Long-course radiotherapy, delayed surgery | Short-course radiotherapy, immediate surgery | Effect |
|---|---|---|---|---|
| Local recurrenceIntention-to-treat analysis (840 randomised)Time to local recurrence (primary endpoint): short-course radiotherapy with delay versus short-course radiotherapy | primaryErlandsson 2017 | — | — | HR 1.44Hazard ratio versus short-course radiotherapy (SRT-delay is the experimental arm in the 2021 report). The 2017 abstract prints the interval as a 95% CI of 0.41-5.11, but it also defines the non-inferiority criterion on a double-sided 90% CI, and the open-access 2021 report states that its local recurrence intervals are 90% CIs; the 2017 full text could not be opened, so the level of the printed interval is unconfirmed. Cumulative incidence of local recurrence in the whole trial: 8 of 357 patients after short-course radiotherapy, 10 of 355 after short-course radiotherapy with delay and 7 of 128 after long-course radiotherapy with delay; p=0.48 is printed once for both comparisons. The abstract states that both delayed regimens were deemed non-inferior, although the upper limits of both printed intervals exceed the 1.7 margin it states; the 2021 report explains that the margin was re-set to 6.5 after the trial began. |
| Local recurrenceIntention-to-treat analysis (840 randomised)Time to local recurrence (primary endpoint): long-course radiotherapy with delay versus short-course radiotherapy | primaryErlandsson 2017 | — | — | HR 2.24Hazard ratio versus short-course radiotherapy; the 2017 abstract prints the interval as 0.71-7.10 without naming the level (see the note on the other comparison). |
| Surgical complicationsThree-arm randomisation (129, 128 and 128 patients)Postoperative complications (three-arm randomisation) | primaryErlandsson 2017 | 38% | 50% | P=0.075Odds ratio versus short-course radiotherapy: 0.59 (95% CI 0.36-0.97) for short-course radiotherapy with delay and 0.63 (0.38-1.04) for long-course radiotherapy with delay. In the pooled comparison of the two short-course regimens the risk of postoperative complications was lower after short-course radiotherapy with delay (odds ratio 0.61, 95% CI 0.45-0.83, p=0.001). The abstract prints the counts as '144 [53%] of 355 vs 188 [41%] of 357'; the percentages do not match the counts (144/355 = 40.6%, 188/357 = 52.7%), and the odds ratio calculated from the counts (0.61) agrees with the printed one, so the percentages appear transposed; only the odds ratio is relied on. Acute radiation-induced toxicity was recorded in 1 of 357 patients (<1%) after short-course radiotherapy, 23 (7%) of 355 after short-course radiotherapy with delay and 6 (5%) of 128 after long-course radiotherapy with delay. |
| OSStockholm III randomised patients (357, 355 and 128 in the whole trial); the abstract does not say which analysis the three medians come from (the three-arm randomisation, 129, 128 and 128 patients, is the only one that includes all three groups)Median overall survival after short-course radiotherapy, short-course radiotherapy with delay and long-course radiotherapy with delay; the abstract gives 'range' values, not confidence intervals | long term-updateErlandsson 2021 | median 10.3 years | median 8.1 years | In the short-course comparison the median overall survival was 8.1 years after short-course radiotherapy and 10.2 years after short-course radiotherapy with delay. The median follow-up for overall survival is 9.8 years (IQR 7.7-12.6) in the full text; the 5.7 years in the abstract is its overall follow-up figure. In the three-arm randomisation local recurrence occurred in 3 of 129, 4 of 128 and 7 of 128 patients and distant metastases in 31, 38 and 38 patients; in the short-course comparison local recurrence occurred in 11 of 357 and 13 of 355 patients and distant metastases in 88 and 82. No comparison showed a statistically significant difference. Full-text hazard ratios for overall survival with short-course radiotherapy as reference: short-course comparison 0.84 (95% CI 0.66-1.06, P=0.146); three-arm randomisation 0.75 (0.51-1.09, P=0.137) for short-course radiotherapy with delay and 0.88 (0.61-1.29, P=0.512) for long-course radiotherapy with delay. |
| Local recurrenceShort-course radiotherapy comparison (357 and 355 patients), analysed by intention to treatTime to local recurrence: short-course radiotherapy with delay versus short-course radiotherapy (reference) | long term-updatemedian follow-up 5.7 yearsErlandsson 2021 | — | — | HR 1.31P=0.518Interval 0.66-2.61; the Methods state that local recurrence intervals are 90% confidence intervals. The Methods also state that the trial was designed as non-inferiority with a margin of 1.7 for the upper limit of a one-sided 90% confidence interval, and that after the trial started a new power calculation was done because local recurrence was less frequent than expected, so that non-inferiority could be decided at an upper limit of 6.5. In the three-arm randomisation the hazard ratio for short-course radiotherapy with delay was 1.18 (0.33-4.15, P=0.830). |
| Local recurrenceThree-arm randomisation (129, 128 and 128 patients), analysed by intention to treatTime to local recurrence: long-course radiotherapy with delay versus short-course radiotherapy (reference) | long term-updatemedian follow-up 5.7 yearsErlandsson 2021 | — | — | HR 2.23P=0.236Interval 0.70-7.14 (90% confidence interval for local recurrence, per the Methods). |
- Short-course radiotherapy, immediate surgery (control)
- Short-course radiotherapy, delayed surgery (experimental)
- Long-course radiotherapy, delayed surgery (experimental)
Median-only curves assume a constant hazard (exponential) and fit plateaus, as seen with some immunotherapies, poorly.
Data table (modelled values)
| Months | Short-course radiotherapy, immediate surgery | Short-course radiotherapy, delayed surgery | Long-course radiotherapy, delayed surgery |
|---|---|---|---|
| 0 | 100.0% | 100.0% | 100.0% |
| 79 | 56.9% | 64.2% | 64.8% |
| 158 | 32.4% | 41.2% | 41.9% |
| 236 | 18.6% | 26.6% | 27.3% |
| 315 | 10.6% | 17.1% | 17.7% |
Appears in
Sources
Erlandsson J, Holm T, Pettersson D, Berglund Å, Cedermark B, Radu C, et al. Optimal fractionation of preoperative radiotherapy and timing to surgery for rectal cancer (Stockholm III): a multicentre, randomised, non-blinded, phase 3, non-inferiority trial. Lancet Oncol. 2017;18(3):336-346.
Erlandsson J, Fuentes S, Radu C, Frödin JE, Johansson H, Brandberg Y, et al. Radiotherapy regimens for rectal cancer: long-term outcomes and health-related quality of life in the Stockholm III trial. BJS Open. 2021;5(6).