JAK Inhibition Results in Aggressive Cutaneous Squamous Cell Carcinoma through MAPK Pathway Activation and Shapes the Tumor Immune Microenvironment to Create Novel Therapeutic Opportunities

Secondary malignancies are the most feared adverse effects of cancer therapies. In this study, we explored the oncogenic effects of the JAK inhibitor ruxolitinib on the skin. First, we conducted a disproportionality analysis on international pharmacovigilance data and demonstrated a signal of increased reporting of cutaneous squamous cell carcinoma (cSCC) with ruxolitinib (PRR 4.88, IC 2.21). We confirmed this association in our monocentric cohort of 1004 patients with myeloproliferative neoplasms (MPN). In multivariable analysis, treatment with JAK inhibitors (JAKi) (HR=6.52, p<0.001) was independently associated with cSCC. Moreover, patients treated with JAKi developed more aggressive cSCC in comparison to untreated patients. To explore the causal link between ruxolitinib exposure and cSCC, we used a chemically induced cSCC mouse model. Both cSCC incidence (100% vs 33%, p=0.079) and aggressiveness criteria were higher in the ruxolitinib group after six weeks of treatment.

Taking advantage of this in vivo model, we mechanistically characterized the effect of JAK-STAT pathway inhibition on cSCC development. Transcriptomic profiling of ruxolitinib and control tumors revealed an enrichment in gene sets related to the MAPK pathway. ERK phosphorylation level increase was confirmed in cSCC histological samples obtained from ruxolitinib treated patients (n=5) in comparison to spontaneously developed cSCC (p<0.001). We then applied a phospho-proteomic approach to unravel the mechanistic underpinnings of ruxolitinib induced MAPK activation. Alternative MAPK signaling downstream of the kinase COT was pinpointed as the main activator of the pathway in this context. The central role of COT was functionally validated using chemical and genetic impairment. Indeed, downregulation of COT rescued the pro-oncogenic effect of ruxolitinib.

Because both MAPK activation and JAK-STAT inhibition are known to induce an immunosuppressive tumor microenvironment, we further explored the extrinsic effects of ruxolitinib exposure. Intriguingly, T CD8⁺ cell number was increased in ruxolitinib induced cSCC both in vitro (p=0.0286) and in vivo (p=0.0159). Combining single-cell transcriptomics, flow-cytometry, immunohistochemistry and cytokine array approaches, we discovered that this T CD8⁺ cell number increase was subsequent to a decreased cSCC immunosuppressive microenvironment in the context of ruxolitinib exposure. Indeed, we demonstrated that JAK-STAT pathway inhibition reduced both immunosuppressive cytokines and immunosuppressive cellular components in the tumor microenvironment, especially T regulatory cells (Tregs). Foxp3 level decrease, used as a canonical marker for Tregs infiltrate, was confirmed in MPN patients cSCC induced by ruxolitinib (p=0.0079). T CD8⁺ subpopulations analysis revealed the increase of PD1⁺TIM3⁺ CD8⁺ T cells under ruxolitinib (p=0.0286). These cells remained in an exhausted state due to their exposure to tumor cells expressing high levels of PDL1, in line with the MAPK pathway hyper-activation under ruxolitinib treatment.

Tregs reduction and T CD8⁺ infiltration are associated with increased checkpoint inhibitors response; suggesting a potential synergistic effect between JAKi and anti-PD1 treatment. In vivo, ruxolitinib addition improved the overall response rate to anti-PD1 therapy both in ruxolitinib induced cSCC (88% vs 17%), and in spontaneously developed cSCC (75% vs 25%). Furthermore, four MPN patients diagnosed with JAKi-induced aggressive cSCC were treated with a combination of JAKi/anti-PD1 after failure of first-line therapy, resulting in 3 complete responses and 1 partial response.

Finally, to evaluate whether the immunomodulating effects of ruxolitinib could also enhance response to immunotherapy in other cancer types, we used a graft-versus-leukemia MLL-AF9 murine model. Indeed, acute myeloid leukemia evolves within an immunosuppressive bone marrow microenvironment, resulting in low efficacy of immunotherapies. In our in vivo model, ruxolitinib reduced Tregs infiltration resulting in an enhanced graft-versus-leukemia anti-leukemic effect (3.48% vs 16.66%).

In conclusion, we revealed that ruxolitinib acts as a double-edged sword in cSCC, inducing MAPK-driven intrinsic pro-oncogenic effects, while remodeling the immune microenvironment to create new therapeutic opportunities for diverse cancer types.