Table 2.

Sensitivity and specificity analysis, and predictive odds ratio of somatic EGFR mutations (A); and EGFR gene copy number gain (B)

A. Comparison of all studies (S23,S34,S35,S43-49,S51-54,S56,S57,S59-81,S83-88,S90-102,S104) and subgroups for somatic EGFR mutations
Somatic EGFR mutationStudies (mutations/patients)Sensitivity (95% CI)Specificity (95% CI)+LR −LRPredictive odds ratio (95% CI)
Overall59 (1,020/3,101)0.78 (0.74- 0.82)0.86 (0.82-0.89)5.570.2622 (16-31)
TreatmentErlotinib8 (87/466)0.64 (0.50-0.75)0.91 (0.86-0.95)7.110.4018 (7-44)
Gefitinib50 (897/2,505)0.79 (0.76-0.83)0.85 (0.81-0.88)5.270.2521 (15-30)
Response criteriaRECIST49 (869/2,672)0.78 (0.74-0.82)0.86 (0.82-0.90)5.570.2623 (16-32)
WHO6 (106/342)0.79 (0.57-0.91)0.84 (0.74-0.91)4.940.2520 (6-65)
EthnicityAsian35 (700/1,677)0.81 (0.76-0.85)0.81 (0.76-0.85)4.260.2318 (13-26)
White19 (251/1,013)0.77 (0.69-0.84)0.91 (0.85-0.95)8.560.2536 (18-72)
RecruitmentUnselected50 (851/2,668)0.77 (0.73-0.81)0.87 (0.83-0.90)5.920.2623 (16-32)
Selected8 (169/433)0.85 (0.72-0.93)0.77 (0.62-0.88)3.700.1919 (8-50)
Funding sourceCompany sponsored13 (211/820)0.70 (0.61-0.78)0.87 (0.81-0.92)5.380.3416 (10-26)
Nonsponsored46 (809/2,281)0.81 (0.76-0.85)0.86 (0.81-0.89)5.790.2225 (17-38)
Detection methodSEQ44 (692/2,248)0.78 (0.72-0.82)0.88 (0.84-0.91)6.500.2525 (17-38)
HS15 (328/853)0.82 (0.77-0.86)0.80 (0.72-0.86)4.100.2318 (11-31)
B. Comparison of all studies (23,34,35,50,55,56,58,59,70,74,75,78,82,83,88,89,91,93,97,99,103) and subgroups for EGFR gene copy number
EGFR gene copy numberStudies (gene gain/patients)Sensitivity (95% CI)Specificity (95% CI+ LR−LRPredictive odds ratio (95% CI)
Overall21 (542/1,539)0.61 (0.49-0.71)0.71 (0.66-0.76)2.100.554 (2-6)
TreatmentErlotinib5 (144/473)0.62 (0.50-0.73)0.75 (0.71-0.79)2.480.515 (3-8)
Gefitinib15 (376/1,030)0.58 (0.43-0.72)0.70 (0.62-0.77)1.930.603 (2-6)
Response criteriaRECIST20 (535/1,515)0.61 (0.49-0.71)0.71 (0.65-0.76)2.100.554 (2-6)
Other1 (7/24)
EthnicityAsian9 (216/483)0.58 (0.47-0.69)0.64 (0.58-0.69)1.610.662 (1-4)
White10 (291/941)0.66 (0.43-0.83)0.75 (0.65-0.82)2.640.456 (2-14)
RecruitmentUnselected18 (434/1,244)0.60 (0.50-0.70)0.72 (0.66-0.77)2.140.564 (3-6)
Selected3 (108/295)
Funding sourceCompany sponsored8 (253/903)0.51 (0.34-0.68)0.76 (0.70-0.82)2.130.643 (2-7)
Nonsponsored13 (289/636)0.66 (0.54-0.77)0.66 (0.59-0.72)1.940.524 (2-7)
Detection methodQ-PCR5 (102/314)0.43 (0.19-0.71)0.79 (0.66-0.88)2.050.723 (1-6)
CCS13 (355/1,034)0.64 (0.50-0.75)0.67 (0.61-0.72)1.940.544 (2-7)
Other3 (61/115)

NOTE: Studies included per subgroup are extractable from Table 1. Subgroups included Treatment, either single-agent erlotinib or gefitinib (no restriction to dose scheduling), excluding studies that did not allow for stratification between treatments; Response criteria, RECIST or amended versions of RECIST versus WHO; Ethnicity, sample population(s) of Asian origin versus others (Whites); Recruitment, studies with patient selection refer to those with artificially enriched populations for any patient characteristic that has been correlated with the presence of somatic mutations in EGFR (e.g., nonsmokers, females, or adenocarcinomas); Funding source, studies that have acknowledged source funding from a recognized body with a potential conflict of interest have been classified as company sponsored (note: this does not imply the inclusion of any intentional bias in the study); Detection method, SEQ, sequencing-based approaches versus more sensitive techniques (HS) in the case of EGFR mutations and FISH/CISH according to The Colorado Classification System (CCS; 22) gene scoring criteria versus quantitative PC (Q-PCR), and scoring systems of FISH/CISH other than the CCS. Studies (n) included in each particular analysis and overall participant numbers (n). In cases where there were less than or equal to four study populations for the meta-regression analysis, the analysis was not conducted.