BACKGROUND Approximately 75% of objective responses to anti–programmed death 1 (PD-1) therapy in patients with melanoma are durable, lasting for years, but delayed relapses have been noted long after initial objective tumor regression despite continuous therapy. Mechanisms of immune escape in this context are unknown. METHODS We analyzed biopsy samples from paired baseline and relapsing lesions in four patients with metastatic melanoma who had had an initial objective tumor regression in response to anti–PD-1 therapy (pembrolizumab) followed by disease progression months to years later. RESULTS Whole-exome sequencing detected clonal selection and outgrowth of the acquired resistant tumors and, in two of the four patients, revealed resistance-associated loss-of-function mutations in the genes encoding interferon-receptor–associated Janus kinase 1 (JAK1) or Janus kinase 2 (JAK2), concurrent with deletion of the wild-type allele. A truncating mutation in the gene encoding the antigen-presenting protein beta-2-microglobulin (B2M) was identified in a third patient. JAK1 and JAK2 truncating mutations resulted in a lack of response to interferon gamma, including insensitivity to its antiproliferative effects on cancer cells. The B2M truncating mutation led to loss of surface expression of major histocompatibility complex class I. CONCLUSIONS In this study, acquired resistance to PD-1 blockade immunotherapy in patients with melanoma was associated with defects in the pathways involved in interferon-receptor signaling and in antigen presentation. (Funded by the National Institutes of Health and others.)

PD-L1 and PD-L2 are ligands for the PD-1 immune inhibiting checkpoint that can be induced in tumors by interferon exposure, leading to immune evasion. This process is important for immunotherapy based on PD-1 blockade. We examined the specific molecules involved in interferon-induced signaling that regulates PD-L1 and PD-L2 expression in melanoma cells. These studies revealed that the interferon-gamma-JAK1/JAK2-STAT1/STAT2/STAT3-IRF1 axis primarily regulates PD-L1 expression, with IRF1 binding to its promoter. PD-L2 responded equally to interferon beta and gamma and is regulated through both IRF1 and STAT3, which bind to the PD-L2 promoter. Analysis of biopsy specimens from patients with melanoma confirmed interferon signature enrichment and upregulation of gene targets for STAT1/STAT2/STAT3 and IRF1 in anti-PD-1-responding tumors. Therefore, these studies map the signaling pathway of interferon-gamma-inducible PD-1 ligand expression.

Loss of function mutations in JAK1/2 can lead to acquired resistance to anti-programmed death protein 1 (PD-1) therapy. We reasoned they may also be involved in primary resistance to anti-PD-1 therapy. JAK1/2 inactivating mutations were noted in tumor biopsies of 1 of 23 patients with melanoma and in 1 of 16 patients with mismatch repair deficient colon cancer treated with PD-1 blockade. Both cases had a high mutational load but did not respond to anti-PD-1 therapy. Two out of 48 human melanoma cell lines had JAK1/2 mutations, which led to lack of PD-L1 expression upon interferon gamma exposure mediated by inability to signal through the interferon gamma receptor pathway. JAK1/2 loss-of-function alterations in TCGA confer adverse outcomes in patients. We propose that JAK1/2 loss-of-function mutations are a genetic mechanism of lack of reactive PD-L1 expression and response to interferon gamma, leading to primary resistance to PD-1 blockade therapy.

Cell-free circulating tumour DNA (ctDNA) in plasma has been shown to be informative of the genomic alterations present in tumours and has been used to monitor tumour progression and response to treatments. However, patients with brain tumours do not present with or present with low amounts of ctDNA in plasma precluding the genomic characterization of brain cancer through plasma ctDNA. Here we show that ctDNA derived from central nervous system tumours is more abundantly present in the cerebrospinal fluid (CSF) than in plasma. Massively parallel sequencing of CSF ctDNA more comprehensively characterizes the genomic alterations of brain tumours than plasma, allowing the identification of actionable brain tumour somatic mutations. We show that CSF ctDNA levels longitudinally fluctuate in time and follow the changes in brain tumour burden providing biomarkers to monitor brain malignancies. Moreover, CSF ctDNA is shown to facilitate and complement the diagnosis of leptomeningeal carcinomatosis.

In the originally published version of this article, the PD-L1 promoter was mistakenly described to be cloned into the BglII/SacI sites of the pGL3 basic vector instead of the BglII site.