Predicting response to pembrolizumab in metastatic melanoma by a new personalization algorithm

Table 9 Personal predictions of TTP

Pred. TTP	Clinic. TTP
Pred. TTP	0–90 days	90–150 days	150–365 days	No progressive disease during follow-up
0–90 days	6 (11.1%)	0 (0%)	2 (3.7%)	2 (3.7%)
90–150 days	0 (0%)	2 (3.7%)	0 (0%)	3 (5.6%)
150–365 days	0 (0%)	0 (0%)	1 (1.8%)	0 (0%)
No progressive disease during follow-up	4 (7.4%)	2 (3.7%)	2 (3.7%)	30 (55.6%)

Comparison between the TTP derived from model predictions (pred.) of the personalization algorithm, and the clinically measured (clinic.) TTP. Each cell includes the number of cases and percentage from the total number of patients in the cohort (in brackets; N = 54), which satisfy one of the six possible outcomes (see “Methods”). The italicized numbers represent the number of cases for which the algorithm correctly predicted whether disease progression will occur, and correctly predicted the time interval in which it occured. Note that our algorithm predicted no progression during the 1-year follow-up period for 30 out of the 35 patients who had not shown clinical progression during that period (bottom right cell). Cohen’s κ = 0.489

ISSN: 1479-5876