Effect of additional OFS on outcomes in patient subgroups

We performed a subgroup meta-analysis based on HR status and chemotherapy experience. Patients were enrolled regardless of HR status, and their ER status was detected after enrollment in the ZIPP and ABC (OAS) trials. Both SOFT and the E-3193 trial enrolled only patients who were ER-positive, progesterone receptor-positive, or both, and provided DFS data. A fixed-effects model was used because no heterogeneity existed between the studies (P=0.96; I2=0%). The meta-analysis results showed that the combined RR for tamoxifen plus OFS versus tamoxifen alone was 0.87 (0.71–1.06;P=0.16; Figure 3A). The ZIPP trial provided OS data for the ER-positive subgroup. The SOFT and E-3193 trials provided OS data for patients who were ER-positive, progesterone receptor-positive, or both. We extracted OS data for the ER subgroup from the ZIPP trial and OS data from the SOFT and E-3193 trials. No significant between-study heterogeneity was found (P=0.93; I2=0%), so the fixed-effects model was used. The meta-analysis results showed that the combined RR for OS regarding tamoxifen plus OFS versus tamoxifen alone was 0.84 (95% CI 0.66–1.07; P=0.16; Figure 3B). The results showed that addition of OFS to tamoxifen did not improve DFS or OS in the HR-positive subgroup.


Continue Reading

(To view a larger version of Figure 3, click here.)

Adjuvant chemotherapy was not permitted in the E-3193 study. Both the SOFT and ZIPP trials provided OS data for the chemotherapy and non-chemotherapy subgroups. The fixed-effects model was used for patients who were not administered chemotherapy because there was no significant between-study heterogeneity (P=0.20; I2=39%). The meta-analysis results showed that the combined RR for tamoxifen plus OFS versus tamoxifen alone was 1.10 (95% CI 0.80–1.51; P=0.55; Figure 4A). The fixed-effects model was also used for patients who were administered chemotherapy because no significant heterogeneity was found (P=0.48; I2=0%). The combined RR for tamoxifen plus OFS versus tamoxifen alone was 0.76 (95% CI 0.60–0.97; P=0.03; Figure 4B). The results showed that addition of OFS to tamoxifen significantly improved OS in the chemotherapy subgroup, but not in the non-chemotherapy subgroups.

(To view a larger version of Figure 4, click here.)

Both the ABC (OAS) and SOFT studies provided distant recurrence data for the tamoxifen and combined OFS arms. The fixed-effects model was used because of no significant heterogeneity (P=0.70; I2=0%). The combined RR for tamoxifen plus OFS versus tamoxifen alone was 0.97 (95% CI 0.86–1.10; P=0.68; Figure 4C). The results showed that addition of OFS to tamoxifen did not reduce distant recurrence.

Publication bias

There was no evidence of publication bias, as shown by inspection of a funnel plot for DFS (Figure 5A) and OS (Figure 5B). Four studies evaluating DFS yielded Begg and Egger P-values of 0.09 and 0.20, respectively. Four studies evaluating OS yielded Begg and Egger P-values of 0.31 and 0.41, respectively. Moreover, publication bias was not detected in the subgroup meta-analyses.

(To view a larger version of Figure 5, click here.)