**4. Discussion**

Our previous work showed that MBZs anti-tumor effect in glioblastomas and medulloblastomas is caused by multiple different mechanisms, such as the inhibition of microtubule formation and VEGFR2 autophosphorylation [17,18], which was corroborated by other investigators, applied to various preclinical cancer models and ultimately translated into clinical trials for adult and pediatric patients with cancer (NCT03925662, NCT03628079, NCT02644291, NCT01729260, NCT01837862). In the current study, we expanded MBZ's scope of application to chemoprevention, i.e., the use of drugs to reduce the risk of cancer development, in high-risk patients of NF1. NF1 is the most common tumor predisposition syndrome in which the loss of tumor suppressor neurofibromin leads to the activation of the Ras proto-oncogene and the development of dozens of benign and malignant tumors. MPNSTs and gliomas are the most common NF1-specific cancers, accounting for 63% of malignancies and a substantial mortality burden in adults younger than 40 years of age; other sarcomas (e.g., rhabdomyosarcomas), gastrointestinal stromal tumors, pheochromocytomas and breast cancers may also occur at a higher frequency compared to the non-affected population [3,4]. MPNSTs in NF1 patients have been particularly recalcitrant to treatment, with overall survival times that are shorter than those of patients with spontaneous MPNSTs. Surgical removal of a high-risk, pre-cancerous lesion is the only prophylactic modality that may reduce mortality, but has unfortunately been associated with morbidity.

In this study, we found that MBZ inhibited the growth of NF1-related MPNST cells in vitro and substantially delayed tumor formation in NPcis mice when initiated 60 days after birth, without overt disease. Interestingly, the effect was different between genders, with male mice experiencing a more substantial protective effect than female mice, who tend to develop tumors later than their male counterparts and have a longer median survival. A similar observation was made

in mice treated with combined CXB with MBZ, which resulted in the largest delay in tumor occurrence and superior median survival in female mice, while males did not experience any benefit from the combination therapy compared to single agen<sup>t</sup> MBZ or CXB. These potential gender-linked distinctions of cancer preventatives are important to realize, as this could impact the clinical applicability of such agents and patient management. Notably, the male bias of NPcis mice in developing MPNST has been reported before [27]. One could also envision intrinsic factors such as the tumor microenvironment, inflammation and di fferences in the sex hormones as potential causes of this phenomenon [26,28,29]; however, the underlying mechanism is unclear and should be investigated further in animals and humans. Our data further sugges<sup>t</sup> that MBZs cytotoxic e ffect on NF1-related malignancies may result from a reduction in activated GTP-bound Ras and a subsequent decrease in pERK in MBZ-treated malignancies in vivo, thus directly targeting the molecular underpinnings for tumor development in this condition. The potential significant impact of such chemoprevention on the mortality rate of cancer in the NF1 patient population can be envisioned from the success of NSAIDs and other agents on reducing the risk of colorectal, prostate and breast cancer. We therefore hope that, by demonstrating the feasibility of a chemopreventative approach for NF1, this will stimulate a rational approach to interrogate already existing databases for drugs that appear to decrease Ras activity and/or increase NF1 expression as a preventative drug discovery pipeline in these patients in order to reduce cancer occurrence and mortality.

Chemoprevention may involve the perturbation of a variety of steps in tumor initiation, promotion and progression. As such, COX-2 overexpression leads to cancer cell proliferation, neovascularization, and suppression of apoptosis and thus is associated with a worse prognosis in various malignancies, especially sarcomas [14–16]. It is therefore not surprising that overexpression of COX-2 has also been observed in NF1-associated MPNSTs and that selective COX-2 inhibition had an antitumor e ffect on these cells [30]. Our study confirmed these results and showed that the selective COX-2 inhibitor CXB, but not the non-selective COX inhibitor SUL, delayed cancer occurrence and increased median survival in both male and female NPcis mice.

E ffective chemoprevention requires the need to identify a high-risk patient population and compounds or drug combinations with very low toxicity to allow long-term use in humans. When initiated 60 days after birth, long-term daily continuous MBZ administration was well tolerated in male and female NPcis mice, with stable weights using 195 mg/kg MBZ feed. This is in line with human data, which demonstrate a >40-year history of safe and continuous use for parasitic infections and cystic echinococcosis. This, along with the observed Ras inhibitory e ffect, could make MBZ an attractive candidate for long-term chemoprevention in the NF1 patient population. It should be noted that rigorous monitoring for adverse reactions would be required, as unexpected and expected toxicities could develop from long-term use of cancer preventative agents, particularly when multiple agents are used, and the benefits should clearly outweigh any potential risks. Given the heterogeneity of clinical symptoms among NF1 patients, it is doubtful that all NF1 patients would experience the same benefits and patient groups at high or low risk would have to be defined. For example, the low-risk NF1 population would include individuals with NF1 Arg1809, NF1 Arg1038Gly, NF1 Met992del, and NF1 Met1149 mutations, all of which are known not to develop any tumors or malignancies [31–34]. In contrast, the largest benefit would likely be observed in patients with a severe phenotype characterized by a higher tumor burden and a higher risk of malignancies. This group of patients would include individuals with large *Nf1* microdeletions, in which the lifetime risk for MPNST is increased to 16–26% [7,35]; patients with an NF1 p.844–848 missense mutation, who have a higher predisposition for symptomatic neurofibromas, optic pathway gliomas and malignancies compared with the general NF1-a ffected population [36] and NF1 patients with Arg1276 variants, who are also at a higher risk of developing symptomatic tumors and MPNSTs [31].

In summary, this study lays an important foundation for the e ffective and feasible chemoprevention of malignancies in patients with NF1, which has the potential to delay or perhaps even prevent the malignant transformation of MPNST and other NF1-related malignancies, decrease the need for surgical intervention and reduce the use of antineoplastic therapies in this patient population. Further research is necessary to evaluate these findings in a larger animal studies, such as the NF1 pig model, and to determine whether the observed effects will result in improved clinical outcomes.

**Author Contributions:** R.-Y.B. and V.S. designed and performed experiments, analyzed data and wrote the manuscript. T.G.-B. provided technical assistance. G.J.R. participated in scientific discussion. All authors have read and agreed to the published version of the manuscript.

**Funding:** V.S. and R.-Y.B. were supported by the Childrens Tumor Foundation Drug Discovery Initiative (DDI) Award and the Department of Defense (DOD) W81XWH1810236. V.S. was also supported by NCI 5K08CA230179.

**Conflicts of Interest:** A patent on MBZ has been filed by JHU, with R-Y.B, V.S. and G.J.R. listed as inventors. The terms of these arrangements are managed by the Johns Hopkins University. In addition, V.S. serves as scientific advisor of the Gilbert Family Foundation—Gene Therapy Initiative outside the submitted work. T.G-B declares no conflict of interest.
