Table 2.

Quantifying treatment benefit (TB) in marker-specific subgroups: examples and clinical interpretation

Trial 1: KRAS is predictive of panitumumab benefit in colon patients (data from ref. 1)
MeasureMarker negative (KRAS mutant)Marker positive (KRAS wild-type)Differential treatment (panitumumab) benefitInterpretation
HR (log scale); PFS0.990.45Log (0.99/0.45; P < 0.0001)There is significant interaction of panitumumab benefit over supportive care and KRAS status over time. In the mutant group, the hazard of progressing is similar (0.99 is close to 1) in patients treated with panitumumab vs. supportive care. In the wild-type group, there is a 55% (1–0.45) decrease in the hazard of progressing in patients treated with panitumumab compared with the hazard of progressing in patients treated with supportive care.
At 6 weeks
Ratio of PFS estimates at 6 weeks1.43 (93/65)1.28 (88/69)1.12 (a)At 6 weeks, PFS was improved in both mutant and wild-type patients; panitumumab benefit for patients with KRAS mutations is 12% better compared with the benefit in wild-type KRAS patients.
Difference in proportion of patients progression-free at 6 weeks28% (93%–65%)19% (88%–69%)9% (a)At 6 weeks, PFS was improved in both mutant and wild-type patients; panitumumab benefit is 9% higher in mutant patients than in wild-type KRAS patients.
At 12 wks
Ratio of PFS estimates at 12 weeks0.57 (8/14)3.33 (50/15)0.17 (a)At 12 weeks, panitumumab benefit for patients with KRAS mutations is 83% worse relative to panitumumab benefit in patients with wild-type KRAS
Difference in proportion of patients progression free at 12 weeks−6% (8%-14%)35% (50%–15%)−41% (a)At 12 weeks, panitumumab benefit is 41% smaller in mutant patients than in wild-type KRAS
Trial 2: PR and PFT treatment in breast patients
PR, progesterone receptor; PFT, L-phenylalanine mustard, 5-fluorouracil, and tamoxifen (datab from ref. 8; subset of women ages <50)
MeasureMarker negative (PR < 10)Marker positive (PR ≥ 10)Differential treatment (PFT) benefitInterpretation
HR (log scale); disease-free-survival (DFS)NRNR−0.52 (P = 0.014)The difference between treatment effects in the PR < 10 and ≥ 10 groups on the HR scale is −0.523 units, regardless of the time point. TB varies according to PR status significantly.
DFS ratio (3 y)0.731.070.68 (P = 0.035)At 3 years, the benefit of PFT in women with PR < 10 is 32% worse relative to the benefit of PFT in women with PR > 10, and it is statistically significant.
Difference in proportion of patients disease-free at 3 years−16%5%−21% (P = 0.035)At 3 years, among women with PR < 10, disease free survival associated with PFT (i.e., PFT benefit) is worse than PF benefit by 16%. PFT benefit among women with PR > 10 is 5% greater than PF benefit. The ATB is 21% and there is a statistically significant difference in TB according to PR status.
  • Abbreviation: NR, not reported.

  • aP values are not provided in examples where raw data were not available to estimate the SE.

  • bDetails for the source of estimates of Trial 2 are shown in Supplementary Table S1.