The application of immune checkpoint inhibitors (ICIs) targeting PD-1/PD-L1 and CTLA-4 has revolutionized cancer treatment in the past decade [1]. However, it is still difficult to determine which patients should be offered immunotherapy currently, and reliable biomarkers are needed [1, 2]. Mutations of NTRK genes are frequently detected in various tumors. They can trigger a number of signal pathways that regulate cell growth, proliferation, differentiation, apoptosis and survival [3], which may impact the tumor immunogenicity. Indeed, previous studies revealed that colorectal tumors harboring NTRK fusions defined a unique subtype with high microsatellite instability [4]. In lung cancer, NTRK alteration was positively associated high tumor mutation burden (TMB) [5]. We speculated the mutation of NTRK could enhance the immune responses and be a potential biomarker in immunotherapy. Therefore, here we conducted a comprehensive bioinformatic and clinical analysis to examine the characteristics of NTRK (NTRK1, NTRK2, and NTRK3) gene mutations and their association with the clinical outcomes of pan-cancer immunotherapy (Suppl. Methods).

Totally, 3888 patients from 14 datasets were included to examine the association between NTRK mutation and the efficacy of immunotherapy (Suppl. Table 1). The discovery cohort was an independent dataset enrolled 1610 patients with 10 cancer types, including lung cancer (n = 344), melanoma (n = 314), bladder urothelial cancer (n = 211), renal cancer (n = 143), head and neck cancer (n = 129), esophagogastric cancer (n = 118), glioma (n = 116), colorectal cancer (n = 109), cancer of unknown primary (n = 85), breast cancer (n = 41). Compared with NTRK non-mutation, patients with NTRK-mutated tumors achieved favorable OS (HR = 0.63; 95% CI, 0.50–0.80; P < 0.001; Fig. 1A). 2278 patients with 7 tumor types from 13 datasets were pooled into the validation cohort. These patients were diagnosed as lung cancer (n = 902), renal cancer (n = 760), melanoma (n = 575), bladder urothelial cancer (n = 27), head and neck cancer (n = 12), sarcoma (n = 1), and anal cancer (n = 1). NTRK mutation was also associated with longer OS (HR = 0.80; 95% CI, 0.65–0.97; P = 0.03; Fig. 1B). Overall, in 3888 patients with 12 cancer types who were treated with ICIs, NTRK mutation (n = 465) decreased the risk of death by 29% (HR = 0.71; 95% CI, 0.61–0.82; P < 0.001; Fig. 1C). Additionally, patients with NTRK mutation showed better ORR (41.7% vs. 27.5%; P < 0.001; Fig. 1D) and PFS (HR = 0.80; 95% CI, 0.68–0.96; P = 0.01; Fig. 1E). Specifically, NTRK3 mutations were discovered in 229 patients and associated with robust anti-cancer activities in terms of ORR (45.4% vs. 28.3%; P < 0.001), PFS (HR = 0.72; 95% CI, 0.58–0.89; P = 0.01), and OS (HR = 0.60; 95% CI, 0.50–0.73; P < 0.001) (Suppl Fig. 1). NTRK2 mutation (n = 105) predicted similar outcomes but to a lesser extent in ORR (40.7% vs. 29.1%; P = 0.05), PFS (HR = 0.70; 95% CI, 0.51–0.95; P = 0.05), and OS (HR = 0.58; 95% CI, 0.45–0.76; P = 0.001). Of note, the predive performances of NTRK1 mutation (n = 188) were only marginal in ORR (38.9% vs. 28.9%; P = 0.02), PFS (HR = 0.87; 95% CI, 0.69–1.09; P = 0.25), and OS (HR = 0.80; 95% CI, 0.63–1.02; P = 0.09).

Both univariate (Fig. 1F) and multivariate (Fig. 1G) Cox analysis confirmed that NTRK mutation was an independent biomarker for OS (HR = 0.83; 95% CI, 0.69–0.99; P = 0.04) and PFS (HR = 0.77; 95% CI, 0.65–0.92; P = 0.004) (Suppl Fig. 2). Hence, we developed a nomogram to estimate 12-month and 24-month OS after the initiation of immunotherapy based on the discovery cohort (Fig. 1H). Further analysis on the calibrations of these predictions suggested this cure-model-based nomogram was good (Suppl Fig. 3). The optimal cutoff value (total points = 130) determined by X-tile software was introduced and categorized patients into high-score and low-score subgroups. Low-score was associated with favorable OS in both discovery cohort (HR = 0.48; 95% CI, 0.41–0.55; P < 0.001; Fig. 1I) and validation cohort (HR = 0.76; 95% CI, 0.64–0.90; P = 0.001; Fig. 1J).

To explore the underlying mechanisms between NTRK mutation and cancer immunotherapy, multi-omics information extracted from the cancer genome atlas (TCGA) cohort were investigated to reveal the tumor immune microenvironment. We first explored the somatic mutant frequencies of three NTRK genes in TCGA pan-cancer cohort. 568 of all 10,953 enrolled patients (5.19%) harbored NTRK mutations. They were found in a small subset of most types of tumors (Suppl. Figure 4), and the mutant frequencies differed significantly among various tumors (P < 0.001). Specifically, NTRK3 mutations were observed in 292 patients (2.67%), NTRK1 in 187 patients (1.71%) and NTRK2 in 170 patients (1.55%). Totally, 733 NTRK mutations were identified (Suppl. Table 1), 606 (82.7%) were missense mutations, 49 (6.7%) were truncating mutations, 38 (5.2%) were spice mutations, 38 (5.2%) were fusion mutations, and 2 (0.3%) were inframe mutations. Moreover, the prognosis for cancer patients were independent of NTRK mutations in terms of OS (HR = 1.09; 95% CI, 0.94–1.27; P = 0.23) and PFS (HR = 1.06; 95% CI, 0.92–1.22; P = 0.45) (Suppl. Figure 5).

The major intrinsic immune response included high tumor immunogenicity, activation of the antigen-processing machinery, and the over-expression of costimulatory molecules [6]. As shown in Fig. 2A, NTRK mutation was associated with higher TMB, non-silent mutation rate, and silent mutation rate. Next, we examined if there were any specific mutation patterns which were associated with the efficacy of immunotherapy. The frequencies of all known COSMIC reference signatures in NTRK-mutant and NTRK-non-mutant tumors were compared. As shown in Suppl Fig. 6A, the frequencies of SBS7a (known etiology, ultraviolet light exposure), SBS10b (POLE mutation), SBS30 (defective DNA base excision repair), and SBS86 (unknown chemotherapy treatment) changed significantly in NTRK-mutant tumors. Further analysis revealed these four signatures were predictive biomarkers for OS in patients treated with ICIs (Suppl Fig. 6B). Additionally, the mRNA expression levels of three major immune checkpoints (PD-1, PD-L1, and CTLA-4) were significantly elevated in NTRK-mutant tumors (Fig. 2B). We also observed most of 16 major histocompatibility complex (MHC) and 25 costimulatory molecules were increased in NTRK-mutant tumors (Fig. 2G).

The key extrinsic immune characteristics included the infiltration of immune cells into the tumor microenvironment, high diversity of B cell receptors (BCRs) and T cell receptors (TCRs), activated immunogenicity of cancer cells contribute to the immune response, and high expression level of immune-stimulators and chemokines [7]. Compared with NTRK-non-mutant tumors (Fig. 2C), NTRK-mutant tumors exhibited higher levels of immune cell infiltration according to (1) leukocyte fractions measured by DNA methylation arrays; (2) lymphocytes fraction estimated from CIBERSORT algorithm; and (3) genomic evaluation of the tumor-infiltrating lymphocyte (TIL) fraction. The abundances of SNV/Indel neoantigens and the diversity of TCR/BCR were significantly upregulated in NTRK-mutant tumors (Fig. 2D). ssGSEA could quantify 29 common immune signatures including key immune pathways, cells, and functions in tumor microenvironment (Fig. 2E) [8]. The MCP-counter method calculated the abundance of 9 immune and 2 stromal cell populations (Fig. 2F) [9]. The immune signatures and cell populations were clearly enriched in NTRK-mutant tumors. Additionally, NTRK-mutant tumors were associated with increased expression of 48 known chemokines and their receptors (Fig. 2H) and 39 immune-stimulators (Fig. 2I).

These results derived from intrinsic and extrinsic immune landscapes indicated that NTRK mutation was associated with enhanced tumor immunogenicity, enriched infiltration of immune cells, and improved immune responses, which might explain that patients with NTRK mutant tumors showed favorable outcomes when treated with ICIs.

In summary, NTRK-mutant tumors might be regarded as immunologically “hot” tumors as they could promote both intrinsic and extrinsic tumor immunogenicity. Moreover, NTRK mutation was an independent biomarker for favorable outcomes in cancer immunotherapy. These results have implications for treatment decision-making and developing immunotherapy for personalized care.