Lesion-Level Response Dynamics to Programmed Cell Death Protein (PD-1) Blockade

Abstract

Purpose: Response to programmed cell death protein 1 (PD-1) blockade is often conceptualized as resulting from reinvigoration of tumor-infiltrating lymphocytes. However, recruited antitumor immunity from the periphery may also be an important contributor to response. A detailed assessment of the response dynamics of individual metastasis could provide insight to the systemic and local features that mediate response and resistance to immunotherapy.

Materials and methods: Patients with metastatic non-small-cell lung cancer (NSCLC) or mismatch repair deficiency (MMRD) carcinoma treated with PD-1 monotherapy were evaluated independently. Absolute and percent change of each target lesion were quantified at each computed tomography scan using RECIST. Patterns of progression were predefined as systemic or mixed and were correlated with clinical outcomes.

Results: A total of 761 individual lesions from 214 patients with NSCLC and 290 lesions from 78 patients with MMRD carcinoma were examined. Individual target lesion responses aligned with best overall response of each patient (85% NSCLC and 93% MMRD lesions responded in patients with partial response/complete response). In responding patients, timing of response was uniform (73% NSCLC and 76% MMRD lesions responded synchronously), and deeper responses were associated with prolonged progression-free survival and overall survival. By contrast, at progression, mixed progression was common (45% of NSCLC and 53% of MMRD) and associated with improved survival compared with those who experienced systemic progression (NSCLC hazard ratio [HR], 0.58; P = .001; MMRD HR, 0.40; P = .07). Organ sites had differential responses, with lymph node and liver metastasis among the most and least responsive, respectively.

Conclusion: Temporal-spatial patterns of response across individual metastases tend to be uniform, favoring the role of peripheral, clonally directed antitumor immunity as a key mediator of response to PD-1 blockade. In contrast, progression is more heterogeneous, potentially revealing the clinical importance of local features and intertumoral heterogeneity.

FIG 1.

Patterns of response in patients treated with programmed cell death protein 1 (PD-1) blockade. (A-F) Spider plot of tumor burden changes of target lesions during the course of treatment with PD-1 blockade. Tumor burden was considered 0 at baseline, and longitudinal follow-up (x-axis) was captured in percentage of change (y-axis). (A) Target lesions (n = 107) from patients with non–small-cell lung cancer (NSCLC) who achieved complete response (CR) or partial response (PR) as their best overall response (BOR) per RECIST (n = 46). (B) Target lesions (n = 159) from patients with NSCLC and stable disease (SD; n = 74). (C) Target lesions (n = 235) from patients with NSCLC and progression of the disease (PD; n = 94). (D) Target lesions (n = 101) from patients with mismatch repair deficiency (MMRD) carcinoma with CR/PR (n = 46). (E) Target lesions (n = 57) from patients with MMRD carcinoma and SD (n = 20). (F) Target lesions (n = 29) from patients with MMRD carcinoma and PD (n = 12). (G-H) Violin plots of the timing of response of individual target lesions in patients that achieve CR or PR per RECIST. The first time at least one target lesion responded (> 30% response) was considered 0; then other target lesions were assessed for relative timing of initial response on the same or subsequent scans. (G) Time to achieve response in target lesions from patients who achieved CR or PR with NSCLC (n = 107). (H) Time to achieve response in target lesions from patients who achieved CR or PR with MMRD carcinomas (n = 101).

FIG 2.

Depth of response and association with clinical outcomes. (A-D) Patients with complete response (CR) or partial response (PR) by RECIST in the non–small-cell lung cancer (NSCLC) cohort (n = 46) and the mismatch repair deficiency (MMRD) cohort (n = 46) were divided into two groups on the basis of the median best overall response (BOR; NSCLC cohort, ≤ v ≥ −51%; and MMRD cohort, ≤ v ≥ −72%, respectively). (A) Progression-free survival (PFS) in the NSCLC cohort, and (B) MMRD cohort comparing these two groups. (C) Overall survival (OS) in the NSCLC, and (D) MMRD cohort comparing these two groups.

FIG 3.

Patterns of progression of disease (PD) during programmed cell death protein 1 (PD-1) blockade. (A-C) Patterns of progression in patients with non–small-cell lung cancer (NSCLC) on the basis of prespecified criteria. (A) Proportion of patients who experienced systemic or mixed progression. Median time (months) and range of progression are described. (B) Kaplan-Meier plots of progression-free survival (PFS), and (C) overall survival (OS) in patients with systemic (n = 94) or mixed progression patterns (n = 78) in the NSCLC cohort. (D-F) Patterns of progression in patients with mismatch repair deficiency (MMRD) carcinoma on the basis of prespecified criteria. (D) Proportion of patients who experienced systemic or mixed progression. (E) Kaplan-Meier plots of PFS, and (F) OS in patients with systemic (n = 14) or mixed progression patterns (n = 16) in the MMRD cohort.

FIG 4.

Site of disease and response to programmed cell death protein 1 (PD-1) blockade. Distribution of best response (%) of individual target lesions to PD-1 blockade grouped by organ of metastasis in patients with (A) non–small-cell lung cancer (NSCLC), and (B) mismatch repair deficiency (MMRD) carcinoma. Response by RECIST criteria, median and range of the percentage of response, and interquartile range of best responses are depicted. Comparisons among responses: *P < .05; ^†P < .01. CR, complete response; PD, progression of disease; PR, partial response; SD, stable disease.

FIG 5.

Hypothesized model of patterns of response and progression to programmed cell death protein 1 (PD-1) blockade. PD-1 blockade is proposed to achieve benefit largely from a systemic, clonal-directed antitumor immune response, which is recruited from the periphery to infiltrate tumors (lavender arrows), consistent with the uniform spatiotemporal pattern of response across metastases seen within a given patient (top right panel). In contrast, patterns of progression can be more mixed, which may reflect the heterogenous and organ-specific properties of the local tumor microenvironment (red dashed circles) that can block or restrain the systemic immune response to PD-1 blockade, and produce mixed progression across metastases within a given patient (bottom right panel).