OR/HR (95% CI) or correlation coefficient; p value | ||

Primary analysis (median follow-up 59 weeks) | Longer-term analysis (median follow-up 108 weeks) | |

Association between log_{2}(baseline CD8^{+} T-cell density) and efficacy endpoints | ||

Baseline CD8^{+} and objective response rate*†‡ | 1.11 (0.87 to 1.42); p=0.387 | 1.06 (0.84 to 1.34); p=0.621 |

Baseline CD8^{+} and durable response rate†‡ | 1.40 (0.99 to 1.97); p=0.056 | 1.18 (0.91 to 1.53); p=0.222 |

Baseline CD8^{+} and disease control rate†‡ | 1.09 (0.87 to 1.37); p=0.436 | 1.09 (0.88 to 1.35); p=0.451 |

Baseline CD8^{+} and duration of response§¶ | 0.74 (0.39 to 1.37); p=0.335 | 0.84 (0.56 to 1.27); p=0.415 |

Baseline CD8^{+} and change in tumor burden** | r=0.03 (n=82); p=0.82 | r=0.01 (n=84); p=0.94 |

Baseline CD8^{+} and overall survival‡§ | 1.15 (0.92 to 1.43); p=0.217 | 1.06 (0.89 to 1.25); p=0.539 |

Baseline CD8^{+} and time to treatment failure‡§ | 0.94 (0.82 to 1.08); p=0.399 | 0.94 (0.82 to 1.07); p=0.336 |

Association between change in log_{2}(CD8^{+} T-cell density) and efficacy endpoints | ||

Change in CD8^{+} and objective response rate††† | 0.94 (0.72 to 1.24); p=0.660 | 1.01 (0.78 to 1.31); p=0.962 |

Change in CD8^{+} and durable response rate††† | 0.99 (0.69 to 1.44); p=0.974 | 0.93 (0.68 to 1.25); p=0.612 |

Change in CD8^{+} and disease control rate††† | 1.08 (0.84 to 1.38); p=0.551 | 1.11 (0.87 to 1.41); p=0.386 |

Change in CD8^{+} and duration of response‡‡‡ | 1.28 (0.47 to 3.47); p=0.626 | 1.22 (0.76 to 1.95); p=0.406 |

Change in CD8^{+} and change in tumor burden** | r=−0.18 (n=56); p=0.18 | r=−0.19 (n=60); p=0.14 |

Change in CD8^{+} and overall survival§†† | 0.85 (0.66 to 1.08); p=0.187 | 0.89 (0.75 to 1.06); p=0.188 |

Change in CD8^{+} and time to treatment failure§†† | 0.94 (0.81 to 1.09); p=0.412 | 0.97 (0.85 to 1.11); p=0.680 |

*Primary endpoint of study.

†A logistic regression model for objective response rate and durable response rate was performed for baseline/change from baseline intratumoral CD8

^{+}T-cell density to obtain unadjusted OR.‡Analysis conducted in biomarker analysis set, n=91.

§A Cox proportional-hazards regression model for duration of response, overall survival and time to treatment failure was performed for baseline/change from baseline intratumoral CD8

^{+}T-cell density to obtain unadjusted HR.¶Analysis conducted in responders who had CD8

^{+}density recorded at baseline: primary analysis n=27, longer-term analysis n=28.**Pearson’s correlation coefficient (r) was estimated to assess the relationship between log

_{2}(CD8^{+}T-cell density) or change in log_{2}(CD8^{+}T-cell density) and the maximum decrease in measurable tumor burden.††Analysis conducted in biomarker evaluable analysis set for non-injected lesions: primary analysis n=59, longer-term analysis n=63.

‡‡Analysis conducted in responders in the biomarker analysis set who had CD8

^{+}density recorded at baseline and week 6: primary analysis n=16, longer-term analysis n=17.CI, confidence interval; HR, hazard ratio; OR, odds ratio.