Table 2. Summarised hazard ratios.
Number of studies | Number of patients | Pooled HR (95% CI) | I2 value (%) | P-value | P-value differences a | |
---|---|---|---|---|---|---|
CD3+ | ||||||
Smallest studiesb | 7 | 1039 | 0.48 (0.33–0.69) | 72 | <0.0001 | <0.00001 |
Largest studiesc | 4 | 1985 | 0.76 (0.55–1.07) | 86 | 0.11 | |
Median follow-up <4 years | 5 | 918 | 0.53 (0.36–0.78) | 65 | 0.001 | <0.000001 |
Median follow-up >4 years | 4 | 1808 | 0.83 (0.66–1.04) | 59 | 0.11 | |
Quality score ⩽5 | 4 | 741 | 0.58 (0.30–1.13) | 89 | 0.11 | 0.11 |
Quality score >5 | 7 | 2301 | 0.58 (0.41–0.82) | 86 | 0.002 | |
Tissue microarray | 5 | 2223 | 0.70 (0.51–0.97) | 85 | 0.03 | <0.00001 |
Whole-tissue slides | 6 | 819 | 0.48 (0.31–0.75) | 77 | 0.001 | |
Hotspots scored | 4 | 671 | 0.62 (0.37–1.04) | 79 | 0.07 | 0.73 |
Representative areas scored | 7 | 2371 | 0.55 (0.37–0.83) | 89 | 0.004 | |
Ovarian cancer | 5 | 1053 | 0.53 (0.28–1.01) | 92 | 0.05 | 0.25 |
Colorectal cancer | 3 | 1319 | 0.64 (0.37–1.10) | 84 | 0.11 | |
CD8+ | ||||||
Smallest studiesd | 12 | 2319 | 0.66 (0.53–0.81) | 59 | 0.0001 | 0.003 |
Largest studiese | 4 | 3740 | 0.82 (0.75–0.91) | 0 | <0.0001 | |
Median follow-up <4 years | 6 | 1322 | 0.53 (0.45–0.62) | 0 | <0.00001 | 0.0003 |
Median follow-up >4 years | 9 | 2539 | 0.76 (0.64–0.90) | 48 | 0.002 | |
Quality score ⩽5 | 7 | 2348 | 0.74 (0.61–0.91) | 48 | 0.004 | 0.46 |
Quality score >5 | 9 | 3693 | 0.69 (0.56–0.86) | 70 | 0.0006 | |
Tissue microarray | 9 | 3926 | 0.71 (0.60–0.85) | 61 | 0.0001 | 0.53 |
Whole-tissue slides | 7 | 2133 | 0.72 (0.55–0.95) | 66 | 0.02 | |
Hotspots scored | 5 | 893 | 0.75 (0.59–0.97) | 37 | 0.03 | 0.98 |
Representative areas scored | 11 | 5166 | 0.70 (0.58–0.83) | 69 | <0.0001 | |
Ovarian cancer | 4 | 983 | 0.69 (0.54–0.88) | 49 | 0.003 | 0.45 |
Colorectal cancer | 4 | 2582 | 0.62 (0.44–0.88) | 81 | 0.007 |
Abbreviations: CIs=confidence intervals; HRs=hazard ratios.
All analyses were carried out for overall survival only. Median follow-up was not available for all studies.
P-values for differences between pairs, bold indicating whether there are statistically significant differences between pooled results from, for instance, small and large studies.
n=100–220.
n=302–786.
n=109–359.
n=500–1290.