Immune-Related Tumor Response Dynamics in Melanoma Patients Treated with Pembrolizumab: Identifying Markers for Clinical Outcome and Treatment Decisions

Abstract

Purpose: Characterize tumor burden dynamics during PD-1 inhibitor therapy and investigate the association with overall survival (OS) in advanced melanoma.Experimental Design: The study included 107 advanced melanoma patients treated with pembrolizumab. Tumor burden dynamics were assessed on serial CT scans using irRECIST and were studied for the association with OS.Results: Among 107 patients, 96 patients had measurable tumor burden and 11 had nontarget lesions alone at baseline. In the 96 patients, maximal tumor shrinkage ranged from -100% to 567% (median, -18.5%). Overall response rate was 44% (42/96; 5 immune-related complete responses, 37 immune-related partial responses). Tumor burden remained <20% increase from baseline throughout therapy in 57 patients (55%). Using a 3-month landmark analysis, patients with <20% tumor burden increase from baseline had longer OS than patients with ≥20% increase (12-month OS rate: 82% vs. 53%). In extended Cox models, patients with <20% tumor burden increase during therapy had significantly reduced hazards of death [HR = 0.19; 95% confidence interval (CI), 0.08-0.43; P < 0.0001 univariate; HR = 0.18; 95% CI, 0.08-0.41; P < 0.0001, multivariable]. Four patients (4%) experienced pseudoprogression; 3 patients had target lesion increase with subsequent response, which was noted after confirmed immune-related progressive disease (irPD). One patient without measurable disease progressed with new lesion that subsequently regressed.Conclusions: Tumor burden increase of <20% from the baseline during pembrolizumab therapy was associated with longer OS, proposing a practical marker for treatment decision guides that needs to be prospectively validated. Pseudoprogressors may experience response after confirmed irPD, indicating a limitation of the current strategy for immune-related response evaluations. Evaluations of patients without measurable disease may require further attention. Clin Cancer Res; 23(16); 4671-9. ©2017 AACR.

Fig. 1

A waterfall plot of the tumor burden change of target lesions at best response (%) in reference to the baseline tumor burden in 96 patients with measurable tumor burden. Three patients noted with the asterisks experienced tumor response ≥30% decrease from baseline after experiencing initial tumor burden increase (pseudoprogression). Five patients noted with # had tumor burden increase beyond +200% (range: 211-566%). While 8 patients achieved CR for the target lesions (-100% of baseline), 3 of them had non-target lesions that did not completely respond and thus had PR for the best overall response assessment. Dotted lines at +20% and -30% represent the threshold for progression (PD) and partial response (PR).

Fig. 2

Spider plot of tumor burden changes during pembrolizumab therapy in 96 patients with measurable tumor burden. Using a upper threshold of +20% increase from baseline tumor burden, the patients whose tumor burden stayed below 20% increase of baseline throughout therapy are noted as a distinct group with apparent treatment benefit (n=57; those below the dashed line of +20%).

Fig. 3

Kaplan-Meier estimates of overall survival (OS) of patients dichotomized by tumor burden changes within 3 months of therapy in the conditional landmark analysis. Patients with <20% tumor burden increase from baseline at 3 months of therapy had longer OS than patients with ≥20% increase from baseline within 3 months (12-month OS rate: 82 vs 53%, respectively).

Fig. 4

Pseudoprogressors with measurable tumor burden. A. The spider plot of tumor burden changes of 3 pseudoprogressors. All patients experienced more than 2 consecutive scans confirming PD (colored arrows) over the period longer than 4 weeks, thus meeting criteria for irPD, before achieving response to therapy (black arrows). B. A 38-year-old female with advanced melanoma with pseudoprogression, corresponding to a green line in Fig. 4A. A baseline scan showed a right axillary lymph node measuring 1.7 cm in short axis (i, arrow). The lesion increased in size on the 1^st at 2.7 months (ii) and 2^nd follow-up scans at 4.1 months (iii), demonstrating increase in size of the lesion more than 20% from baseline, confirming irPD. The lesion reached its maximal size at the 3^rd follow-up scan at 5.5 months (iv), and then started to decrease in size on the 4^th scan at 6.7 months (v). The lesion further decreased in size gradually, and met the criteria for response at 22.3 months of therapy (vi). Since then, the lesion remained small and maintained durable response over 19 months.

Fig. 4

Pseudoprogressors with measurable tumor burden. A. The spider plot of tumor burden changes of 3 pseudoprogressors. All patients experienced more than 2 consecutive scans confirming PD (colored arrows) over the period longer than 4 weeks, thus meeting criteria for irPD, before achieving response to therapy (black arrows). B. A 38-year-old female with advanced melanoma with pseudoprogression, corresponding to a green line in Fig. 4A. A baseline scan showed a right axillary lymph node measuring 1.7 cm in short axis (i, arrow). The lesion increased in size on the 1^st at 2.7 months (ii) and 2^nd follow-up scans at 4.1 months (iii), demonstrating increase in size of the lesion more than 20% from baseline, confirming irPD. The lesion reached its maximal size at the 3^rd follow-up scan at 5.5 months (iv), and then started to decrease in size on the 4^th scan at 6.7 months (v). The lesion further decreased in size gradually, and met the criteria for response at 22.3 months of therapy (vi). Since then, the lesion remained small and maintained durable response over 19 months.

Fig. 5

Pseudoprogression in the setting of no measurable tumor burden at baseline. A 66-year-old male with advanced melanoma treated with pembrolizumab. A. The baseline scan showed no measurable tumor burden and subcentimeter, non-measurable brain metastasis. B. A follow-up scan at 1.5 months of therapy showed a new subcutaneous nodule (arrow). C. On a subsequent follow-up scan at 3.5 months of therapy, the nodule has significantly decreased in size (arrow).