Defective mismatch repair as a predictive marker for lack of efficacy of fluorouracil-based adjuvant therapy in colon cancer
Needs reviewPhase observationalN = 457
- Population
- 457 patients with resected stage II or III colon cancer and a mismatch repair result (70 dMMR, 15%); 229 randomised to fluorouracil-based therapy and 228 to surgery alone. Pooled with the earlier analysis: 1027 patients (165 dMMR).
- Design
- Retrospective biomarker analysis of specimens from patients previously randomised in five completed trials (FFCD 8802, NCCTG 78-48-52, NCCTG 87-46-51, Intergroup 0035 and GIVIO) of fluorouracil with levamisole or leucovorin versus surgery alone, in patients not used in the earlier MSI analysis (Ribic 2003); mismatch repair status by MSI testing (GIVIO) or immunohistochemistry for MLH1 and MSH2. These data were then pooled with the earlier analysis. Primary endpoint disease-free survival (censored at 5 years); overall survival secondary. The primary analyses were in the new cohort; by-stage and pooled analyses were secondary. Not a new randomised comparison.
- Primary endpoint
- DFS
- Arms
- Surgery alone (control, n = 228): Surgery alone (control arms of the source trials)
- Fluorouracil-based adjuvant therapy (experimental, n = 229): Fluorouracil with levamisole or leucovorin (source-trial regimens)
- People
- Sargent DJ
DFSHR 0.67 (95% CI 0.48–0.93)DFSHR 1.10 (95% CI 0.42–2.91)DFSHR 1.39 (95% CI 0.46–4.15)OSHR 2.95 (95% CI 1.02–8.54)DFSHR 1.53 (95% CI 0.78–3.04)Δ 60-month DFS−10 points
Show the numbers
| Endpoint | Analysis | Fluorouracil-based adjuvant therapy | Surgery alone | Effect |
|---|---|---|---|---|
| DFSPatients with pMMR tumours in the 457-patient cohort not used in the earlier analysis | exploratorySargent 2010 | — | — | HR 0.67 (95% CI 0.48–0.93)P=0.02Fluorouracil-based therapy versus surgery alone; retrospective analysis of specimens from randomised trials. |
| DFSPatients with dMMR tumours in the 457-patient cohort (70); values as printed in the abstract | exploratorySargent 2010 | — | — | HR 1.10 (95% CI 0.42–2.91)P=0.85Fluorouracil-based therapy versus surgery alone. These are the abstract's values. The full text of the author manuscript gives different values for the same comparison (see the next result, dfs-dmmr-new-cohort-full-text: multivariate HR 1.39, 95% CI 0.46 to 4.15, P=0.56; univariate HR 1.44, 0.55 to 3.81, P=0.46); the source does not explain the difference, and PubMed lists an erratum for the paper (J Clin Oncol 2010;28(30):4664) whose text could not be read. |
| DFSPatients with dMMR tumours in the 457-patient cohort (70: 37 surgery alone, 33 treated); multivariate model, values as printed in the full text (Table 2 and Results) | exploratorySargent 2010 | — | — | HR 1.39 (95% CI 0.46–4.15)P=0.56Fluorouracil-based therapy versus surgery alone, from the full text. The abstract gives HR 1.10 (95% CI 0.42 to 2.91; P=.85) for the same comparison (see dfs-dmmr-new-cohort). Univariate HR in Table 2: 1.44 (0.55 to 3.81), P=.46. Methods state that dMMR models were not stratified and were adjusted for stage only, whereas the Table 2 title says adjusted for stage, sex and age. Table 2 gives 5-year DFS of 76% (surgery alone) and 71% or 72% (treated; the two blocks of the table differ). The interaction test between MMR status and treatment efficacy in this cohort was not significant (P=.18). |
| OSPatients with stage II disease and dMMR tumours in the pooled data set (1027 patients, 165 dMMR) | exploratorySargent 2010 | — | — | HR 2.95 (95% CI 1.02–8.54)P=0.04Treated versus surgery alone; a secondary, by-stage analysis in a small subgroup with a wide confidence interval. The full text states that all overall-survival findings were consistent with the DFS findings in direction and significance, with this one exception. |
| DFSPatients with dMMR tumours in the pooled data set (165: 79 untreated, 86 treated) | exploratorySargent 2010 | 70% at 5 y | 80% at 5 y | HR 1.53 (95% CI 0.78–3.04)P=0.22Multivariate hazard ratio (treated versus untreated), adjusted for stage, sex and age per the Table 4 title (the Methods say models in dMMR tumours were adjusted for stage only); whole-number 5-year rates as printed in Table 4. The interaction test between MMR status and treatment efficacy for DFS in the pooled data set was significant (P=0.04, full text). Treated versus untreated is inferred from the rates (80% untreated, 70% treated, HR above 1); the table does not label the direction. |
| DFSPatients with pMMR tumours in the pooled data set (862: 436 untreated, 426 treated) | exploratorySargent 2010 | 67% at 5 y | 56% at 5 y | HR 0.70 (95% CI 0.56–0.88)P=0.002Multivariate hazard ratio (treated versus untreated), adjusted for stage, sex and age per the Table 4 title (the Methods say models in dMMR tumours were adjusted for stage only); whole-number 5-year rates as printed in Table 4. Treated versus untreated is inferred from the rates (56% untreated, 67% treated, HR below 1); the table does not label the direction. |
| Endpoint | Analysis | Fluorouracil-based adjuvant therapy | Surgery alone | Effect |
|---|---|---|---|---|
| DFSPatients with pMMR tumours in the 457-patient cohort not used in the earlier analysis | exploratorySargent 2010 | — | — | HR 0.67 (95% CI 0.48–0.93)P=0.02Fluorouracil-based therapy versus surgery alone; retrospective analysis of specimens from randomised trials. |
| DFSPatients with dMMR tumours in the 457-patient cohort (70); values as printed in the abstract | exploratorySargent 2010 | — | — | HR 1.10 (95% CI 0.42–2.91)P=0.85Fluorouracil-based therapy versus surgery alone. These are the abstract's values. The full text of the author manuscript gives different values for the same comparison (see the next result, dfs-dmmr-new-cohort-full-text: multivariate HR 1.39, 95% CI 0.46 to 4.15, P=0.56; univariate HR 1.44, 0.55 to 3.81, P=0.46); the source does not explain the difference, and PubMed lists an erratum for the paper (J Clin Oncol 2010;28(30):4664) whose text could not be read. |
| DFSPatients with dMMR tumours in the 457-patient cohort (70: 37 surgery alone, 33 treated); multivariate model, values as printed in the full text (Table 2 and Results) | exploratorySargent 2010 | — | — | HR 1.39 (95% CI 0.46–4.15)P=0.56Fluorouracil-based therapy versus surgery alone, from the full text. The abstract gives HR 1.10 (95% CI 0.42 to 2.91; P=.85) for the same comparison (see dfs-dmmr-new-cohort). Univariate HR in Table 2: 1.44 (0.55 to 3.81), P=.46. Methods state that dMMR models were not stratified and were adjusted for stage only, whereas the Table 2 title says adjusted for stage, sex and age. Table 2 gives 5-year DFS of 76% (surgery alone) and 71% or 72% (treated; the two blocks of the table differ). The interaction test between MMR status and treatment efficacy in this cohort was not significant (P=.18). |
| OSPatients with stage II disease and dMMR tumours in the pooled data set (1027 patients, 165 dMMR) | exploratorySargent 2010 | — | — | HR 2.95 (95% CI 1.02–8.54)P=0.04Treated versus surgery alone; a secondary, by-stage analysis in a small subgroup with a wide confidence interval. The full text states that all overall-survival findings were consistent with the DFS findings in direction and significance, with this one exception. |
| DFSPatients with dMMR tumours in the pooled data set (165: 79 untreated, 86 treated) | exploratorySargent 2010 | 70% at 5 y | 80% at 5 y | HR 1.53 (95% CI 0.78–3.04)P=0.22Multivariate hazard ratio (treated versus untreated), adjusted for stage, sex and age per the Table 4 title (the Methods say models in dMMR tumours were adjusted for stage only); whole-number 5-year rates as printed in Table 4. The interaction test between MMR status and treatment efficacy for DFS in the pooled data set was significant (P=0.04, full text). Treated versus untreated is inferred from the rates (80% untreated, 70% treated, HR above 1); the table does not label the direction. |
| DFSPatients with pMMR tumours in the pooled data set (862: 436 untreated, 426 treated) | exploratorySargent 2010 | 67% at 5 y | 56% at 5 y | HR 0.70 (95% CI 0.56–0.88)P=0.002Multivariate hazard ratio (treated versus untreated), adjusted for stage, sex and age per the Table 4 title (the Methods say models in dMMR tumours were adjusted for stage only); whole-number 5-year rates as printed in Table 4. Treated versus untreated is inferred from the rates (56% untreated, 67% treated, HR below 1); the table does not label the direction. |
Data table
| Endpoint | HR | 95% CI |
|---|---|---|
| DFS | 0.67 | 0.48–0.93 |
| DFS | 1.1 | 0.42–2.91 |
| DFS | 1.39 | 0.46–4.15 |
| OS | 2.95 | 1.02–8.54 |
| DFS | 1.53 | 0.78–3.04 |
| DFS | 0.7 | 0.56–0.88 |