Skip to main content
. 2017 Jun 12;12(6):e0179536. doi: 10.1371/journal.pone.0179536

Table 3. Subgroup analysis and sensitive analysis on the outcome of disease-free survival.

No. of Studies No. of patients Heterogeneity Statistical model used HR (95% CI) Significance
Subtype of HNC
1. HNSCC 2 540 I2 = 79.1%; P = 0.028 Random 0.58 (0.15, 2.30) P = 0.438
2. OSCC 3 525 I2 = 68.5%; P = 0.042 Random 1.73 (0.88, 3.42) P = 0.112
3. OPSCC 1 133 NA Fixed 1.68 (0.75, 3.76) P = 0.206
4. LSCC 1 238 NA Fixed 0.59 (0.35, 1.00) P = 0.048
5. NPC 3 382 I2 = 15.0%; P = 0.308 Fixed 1.77 (1.32, 2.37) P = 0.001
6. SGC 1 219 NA Fixed 2.29 (1.25, 4.19) P = 0.007
PD-L1 expression
1. mRNA 1 517 NA Fixed 1.05 (0.65, 1.45) P = 0.812
2. Protein 10 1520 I2 = 73.8%; P < 0.001 Random 1.48 (0.99, 2.20) P = 0.056
Origin of population
1. Europe 4 868 I2 = 75.6%; P = 0.006 Random 0.84 (0.44, 1.61) P = 0.598
2. Asia 6 952 I2 = 0.0%; P = 0.463 Fixed 1.99 (1.59, 2.48) P = 0.001
Cut-off value for PD-L1 positive
1. > 5% positive expression 3 525 I2 = 68.5%; P = 0.042 Random 1.73 (0.88, 3.42) P = 0.112
2. H score > 35 2 278 I2 = 37.6%; P = 0.206 Fixed 1.71 (1.26, 2.32) P = 0.011
Sample size of studies
1. < 100 2 113 I2 = 88.3%; P = 0.003 Random 0.77 (0.10, 6.15) P = 0.806
2. > 100 9 1924 I2 = 71.7%; P < 0.001 Random 1.52 (1.06, 2.17) P = 0.022
HR estimation
1. Reported 6 1429 I2 = 78.6%; P < 0.001 Random 1.54 (0.91, 2.60) P = 0.106
2. Available data calculated 5 608 I2 = 70.9%; P = 0.008 Random 1.29 (0.73, 2.28) P = 0.386
HR Calculation
1. Multivariate analysis 2 437 I2 = 0.0%; P = 0.719 Fixed 2.50 (1.71, 3.65) P = 0.001
2. Univariate analysis 9 1600 I2 = 70.0%; P = 0.001 Random 1.23 (0.84, 1.82) P = 0.289

HNC, head and neck cancer; HNSCC, head and neck squamous cell carcinoma; OSCC, oral squamous cell carcinoma; OPSCC, oropharyngeal squamous cell carcinoma; LSCC, laryngeal squamous cell carcinoma; NPC, nasopharyngeal carcinoma; SGC, salivary gland carcinoma; PD-L1, programmed cell death ligand 1; HR, hazard ratio; CI, confidence interval; NA, not applicable.