Table 1.

Associations between pretreatment levels of sPD-L1 and clinical responses. A and B, Ipilimumab plus bevacizumab treatment. C and D, Either ipilimumab or ipilimumab plus sargramostim treatment in ECOG 1608 trial. “Unevaluable” indicated that the patients deceased in short time and did not receive evaluation. Statistic results between analyses with and without “unevaluable” were similar. E and F, Anti–PD-1 treatment.

Association between pretreatment levels of sPD-L1all and clinical responses
A. Ipilimumab plus bevacizumab
sPD-L1all of pretreatment (baseline, ng/mL)PRSDPDTotal
<1.48 (21%)21 (57%)8 (22%)37
≥1.40 (0%)0 (0%)5 (100%)5
Fisher exact test P = 0.0015
C. Ipilimumab or ipilimumab plus sargramostim
sPD-L1all of pretreatment (baseline, ng/mL)PRSDPDTotal
<1.432 (23%)32 (23%)77 (54%)141
≥1.40 (0%)1 (10%)9 (90%)10
Fisher exact test P = 0.04
E. Anti–PD-1 antibody
sPD-L1all of pretreatment (baseline, ng/mL)PRSDPDTotal
<1.411 (34%)8 (25%)13 (41%)32
≥1.41 (33%)0 (0%)2 (67%)3
Association between pretreatment levels of sPD-L1L and clinical responses
B. Ipilimumab plus bevacizumab
sPD-L1L of pretreatment (baseline, ng/mL)PRSDPDTotal
<0.58 (20%)21 (54%)10 (26%)39
≥0.50 (0%)0 (0%)3 (100%)3
Fisher exact test P = 0.025
D. Ipilimumab or ipilimumab plus sargramostim
sPD-L1L of pretreatment (baseline, ng/mL)PRSDPDTotal
<0.530 (22%)31 (22%)78 (56%)139
≥0.52 (17%)2 (17%)8 (66%)12
Fisher exact test P = 0.55
F. Anti–PD-1 antibody
sPD-L1L of pretreatment (baseline, ng/mL)PRSDPDTotal
<0.511 (33%)8 (24%)14 (43%)33
≥0.51 (50%)0 (0%)1 (50%)2