1815. Recurrent Melanoma: Imaging Predictors of Immunotherapy (IT) Response and Survival
Authors * Denotes Presenting Author
  1. Milos Janicek *; Boston University Medical Center
  2. Adam Lerner; Boston University Medical Center
  3. Joseph Park; Boston University Medical Center
Aims to predict the effectiveness/duration of response to immunotherapy (IT) measured by overall survival (OS) after initiation of immunotherapy (OSpT) in recurrent melanoma failing original curative attempt (OCA) utilizing new baseline imaging characteristics and iRECIST-derived score (iRs). Identify distinct risk groups based on iRs change on two sequential early restaging scans (eSRS) and concordance/discordance of response.

Materials and Methods:
This study included a total of 38 patients (23 men, 15 women, age 36–95 years) with stage 4 melanoma at start of IT. Up to five lesions were evaluated using iRECIST criteria for interval change; iRs were assigned based on favorable characteristics at new baseline for male sex, oligometastatic, no liver mets, long FFF from OCA (median FFF = 341 d); 1 point for each feature or –1 point for absence of such feature. It was compared with the added value of eSRS at 4 and 7.6 months after start of IT. Baseline iRs were compared to two eSRS for change at ± 10%, categorized as PP, Sx, Rx, PR (i.e., P=progression, R=regression, S=stable, x=any) with point values –2, –1, 1, and 2, respectively. Observation of discordant response was recorded and scored as –1 point, concordant response was 1 point. Cumulative scores were recorded at baseline and at second restaging (with additional benefit of eSRS) and compared for bad = –2, medium = –2 to 2, good = 2. Kaplan–Meyer, K-statistics for the groups were utilized at p < 0.05 for significance.

Using MEDIAN OSpT of 464 days, "Long OSpT" was associated with above listed "favorable characteristic" at the new baseline, yet SCORE did NOT separate prognostic groups significantly. Adding sequential imaging derived score at 2nd restaging scans, SIGN separation of 3 prognostic groups was seen: Using cumulative score of <-2, -2to2, >2, MEAN OSpT was 264d, 546d, 1206d, with SIGN separation at P=.026, P=.0004 and P=.09 respectively). Evaluating eSRS response curves separately, "pseudo progression" was associated with the best OSpT, lack of response (S) was unfavorable and similar to PP on two sequential scans. Discordant response on eSRS was also associated with significantly shorter OSpT, all these SIGN at P<0.05.

New challenges of IT are reshaping traditional staging concepts. Each disease entity may have very specific imaging needs/presentations, in spite of unifying features of IT dynamics. Significant benefit of IT in the responder group is weighted against significant toxicity - separation of these groups is essential. There is equal proportion of patients who could benefit from a shorter course of therapy - some may need maintenance IT. Therefore, separation of these subgroups is of utmost clinical importance and imaging plays an essential role. The study validated the role of sequential imaging, in addition to baseline staging at the time of OCA failure, which alone did not separate favorable/unfavorable prognostic groups with respect to OSpT. There was a significant added value of eSRS categorizing response curves separating the poor prognostic group from medium and favorable ones at p < 0.01.