Gastroenteropancreatic neuroendocrine carcinomas (GEP NECs) are neoplasms that arise from malignant transformation of cells within the diffuse neuroendocrine system of the gastrointestinal tract and pancreas. Well-differentiated neuroendocrine tumors (NETs) generally grow slowly, and may secrete hormones.¹ Their poorly differentiated counterparts, GEP NECs, are rarer, but their incidence has increased over the past 40 years.² These tumors are much more aggressive than NETs, with 5 year overall survival rates of 40% and 7% for high-grade midgut NECs and pancreatic NECs (pNECs), respectively.³ There has been increasing interest in investigating new treatments for these tumors due to these poor outcomes and limited therapeutic options.^{3, 4}

Current studies are looking into the response of GEP NECs to doxorubicin (DOX) and temozolomide (TMZ)-based therapies, and results have been promising.^{5, 6, 7, 8, 9} DOX is a commonly used anti-cancer drug that acts through many mechanisms: generation of reactive oxygen species, topoisomerase II inhibition, histone disassociation, ceramide overproduction, and DNA-adduct formation.¹⁰ DOX given in combination with 5-fluorouracil and streptozocin has been shown to improve progression-free survival (PFS) and overall survival (OS) in well-differentiated pNETs.⁹ TMZ, a newer anti-cancer drug, is an alkylating agent which methylates guanine bases leading to the formation of DNA strand nicks and apoptosis.¹¹ Advanced NETs, including gastric, bronchial, and pancreatic, have shown a good response to TMZ as a monotherapy in previous retrospective and small prospective studies.^{5, 6} A recent multicenter, randomized, phase II trial showed that TMZ combined with capecitabine, a regimen known as CAPTEM, showed statistically and clinically significant increased PFS compared to TMZ alone in patients with advanced pNETs.⁷ Per the most recent NANETs guidelines, TMZ-based treatments are considered a viable systemic treatment for metastatic pheochromocytomas and paragangliomas, well-differentiated and metastatic pNETs, metastatic thymic NETs, and metastatic lung/bronchial NETs. In terms of NECs, TMZ-based regimens are considered a reasonable second-line option for pNECs after failing cisplatin or carboplatin in combination with etoposide.¹² However, data is insufficient to recommend TMZ-based regimens over other second-line options.^{13, 14, 15, 16, 17} Despite these advances in therapies, prognosis of high-grade NECs remain poor and metastatic GEP NECs are considered incurable warranting continued investigation into new effective therapies.^{3, 4}

One limitation to ongoing investigation is the lack of representative GEP NEC models. Recently, two new NEC cell lines, NEC1452 and NEC913 were developed and characterized. These lines grow well both in vitro and in vivo, and allow for expanded investigations of drug sensitivity in GEP NECs.¹⁸ This study aims to identify new categories of therapeutic agents that can target GEP NECs through a systematic screen of the NEC913 and NEC1452 cell lines with an established library of compounds from the National Cancer Institute (NCI) Diversity Set VII collection. Compounds from this collection have different structures and have not previously been screened for activity in GEP NEC cells.

NEC913 and NEC1452 Patient-Derived Xenograft (PDX) models were previously created in our laboratory and are valuable models of GEP NECs for testing therapeutic agents.¹⁸ These NEC PDX tumors can be passaged in immunocompromised mice and harvested for tumor cell isolation (Figure 1A). The NEC913 and NEC1452 tumor cells were found to maintain expression of NEC markers. Chromogranin A (CgA) and synaptophysin (SYP) were expressed in both NEC913 and NEC1452 lines whereas somatostatin receptor 2 (SSTR2) was only expressed in NEC913 (Figure 1B).

To identify potential therapeutic compounds, 885 structurally diverse compounds were systematically screened against both NEC913 and NEC1452 spheroids (Figure 1C). Of the 885 tested, 16 were found to inhibit over 75% of spheroid survival (Table 1 and Figure 2). NEC1452 showed sensitivity to 11 of these compounds whereas NEC913 was sensitive to 6 (Figure 2E). Only 1 compound, 2,2’,5,5’-tetrahydroxy-4,4’-dimethylbiphenyl, showed inhibition in both cell lines. Analysis of these 16 compounds in the PubChem database showed that 7 had a similar target in tyrosyl-DNA phosphodiesterase 1 (TDP1) (Table 1). NEC1452 cells showed sensitivity to all 7 TDP1-targeting compounds, whereas NEC913 cells were sensitive to none. The other 7 inhibitors target epigenetic regulators such as histone methylase and demethylase enzymes. Two compounds have unknown targets (Table 1).

Of the 7 TDP1-targeting compounds, only 2 (echinosporin [ECHINO] and mitomycin C [MitoC]) are commercially available. These were purchased for testing in combination with DOX and TMZ, 2 FDA-approved drugs that have been used in treating neuroendocrine neoplasms. ^{5, 6, 7, 8, 9, 12} In addition, we tested 2 previously reported TDP1 inhibitors, cytarabine (CYT) and neomycin (NEO), in combination with DOX and TMZ.^{19, 20} These TDP1 inhibitor and DOX or TMZ combinations were screened at varying concentrations for a synergistic effect on NEC1452 cells. CYT showed an additive effect on inhibition when combined with DOX as well as TMZ (Figure 3) with a ZIP synergy score of 5.585 and 8.949, respectively (Figure 3A&B). At 0.09 μM, DOX alone significantly decreased relative growth to 0.87 when compared with control (p<0.01) (Figure 3C). The inhibitory effect of DOX was enhanced when combined with CYT at all tested concentrations. Relative growth of NEC1452 cells at 3.75 μM TMZ was significantly decreased to 0.92 compared to control (p<0.01) (Figure 3D). When 3.75 μM TMZ was combined with CYT at all tested concentrations, there was a significant decrease in relative growth compared to TMZ alone (p<0.05) (Figure 3D).

To further investigate the biochemical effects of combination therapy on NEC1452 cells, a western blot was performed to probe for markers of double-stranded DNA breaks (γH2AX) and cell proliferation (phosphorylated-ERK). The combination of 1.5 μM CYT to 15 μM TMZ showed an increase in γH2AX (Figure 4A) compared to TMZ alone after 18 hours of treatment and a decrease in phosphorylated-ERK (phospho-ERK) (Figure 4B) compared to TMZ alone after 24 hours of treatment, indicating increased double-stranded breaks and decreased cell proliferation. TMZ with ECHINO as well as TMZ and NEO showed no significant differences in phosphorylated-ERK or γH2AX compared to TMZ alone (Figure 4).

GEP NECs are rare neoplasms with few systemic treatment options and poor prognosis.^{3, 4} Here, we have shown that NEC1452 cells are sensitive to TDP1 inhibitors, a class of drugs not currently used in NEC treatments. Historically, TDP1 has been considered a target for anticancer therapies; however, few TDP1 inhibitors exist.²¹ Our drug screen identified 2 commercially available TDP1 inhibitors (echinosporin and Mitomycin C). We included 2 additional TDP1 inhibitors (cytarabine and neomycin) in our study as well.^{20, 21, 22, 23, 24} Cytarabine, although not a true inhibitor of TDP1, acts within the TDP1 pathway through inhibition of TOP1 leading to increased TOP1cc (Figure 5). Here, we have shown that cytarabine enhances the cytotoxic effect of TMZ in our NEC1452 cells. TMZ and TMZ-based regimens are associated with significant clinical responses in pancreatic NETs and metastatic NETs, but the data in GEP NECs is not as robust.^{5, 6, 7, 8, 13, 14, 17, 25, 26} Thus, the ability of TDP1 pathway inhibitors to increase sensitivity to TMZ in NEC1452 cells, a line derived from a rectal primary NEC, is significant. This opens a new avenue of possibilities with TMZ-based regimens in combination with TDP1 inhibitors within the NEC sector that is worthy of pursuing further. Additional studies looking at other combinations with TDP1 inhibitors, like the common pNEC therapies etoposide and cisplatin, could further define TDP1 inhibitors potential role in NEC therapy.¹²

Interestingly, echinosporin and neomycin, which are more direct TDP1 inhibitors, showed no change in γH2AX or phosphorylated-ERK when combined with TMZ (Figure 4). This was likely due to the small concentrations of the drugs used in this study (1.5 μM). Previous studies have shown that neomycin is only active against TDP1 at high concentrations, with one report showing a half maximal inhibitory concentration of 8 mM.²⁷ Thus, neomycin, and likely echinosporin, are only weak inhibitors of TDP1. At the necessary levels for TDP1 inhibition, neomycin is known to be ototoxic and nephrotoxic.^{28, 29} Furthermore, neomycin is poorly absorbed orally and ineffective systemically making it an unsuitable option for a combination chemotherapy regimen with TMZ.¹⁹ Unlike neomycin, cytarabine is a current anticancer drug used to treat acute myelogenous leukemia and lymphocytic leukemias and therefore known to be clinically tolerable.^{30, 31}

Although the compounds from the NCI Diversity Set VII provide us with many structurally diverse compounds for screening, the main limitation of using this library is that most of the compounds are not FDA-approved drugs. Hence, research into finding an FDA-approved drug acting on the same target is required to achieve a translational goal. In addition, often higher concentrations of these compounds are required to obtain growth inhibition as we have observed for neomycin and echinosporin. We used a relatively high dose (3 μM) for our screen and only found 16/885 compounds (equivalent to 2%) have inhibitory effects on NEC spheroids. We predict that using a higher concentration during the drug screen could increase the number of active compounds identified.

Finally, there was a notable difference in sensitivity to TDP1 inhibition between NEC1452 and NEC913 in our original drug screen. All 7 TDP1 inhibitors were found to decrease growth of NEC1452 cells over 75%, and none reached this level of inhibition in NEC913 cells. The difference in sensitivity to TDP1 inhibitors suggests a difference in amount of TDP1 in each cell line. NEC913 cells maintain expression of SSTR2 and tend to grow at slower rates. In contrast, NEC1452 has lost expression of SSTR2 and grow more quickly. As NEC913 go through fewer rounds of replication within a given treatment period, they have more time for DNA repair machinery to take action compared to NEC1452, potentially decreasing the amount of TOP1cc through mechanisms independent of TDP1 (Figure 5).²⁴ In addition, the NEC1452 cells have the APC gene mutated which render them less efficient at base excision repair. With less TOP1cc to cleave, TDP1 may play a more minor role seen as decreased transcription and translation of the TDP1 gene. Thus, inhibition of TDP1 in NEC913 cells may not lead to a significant decrease in growth. Further testing of NEC1452 and NEC913 cells to quantify levels of TDP1 would be useful to determine the correlation with response to these agents for NETs and NECs.

In conclusion, TDP1 inhibitors are a novel class of drugs found to inhibit growth of one GEP NEC cell line. They were also found to work synergistically with TMZ to decrease NEC cell growth and increase double-stranded breaks compared to TMZ alone. As there remain few systemic options for GEP NECs, these findings reveal new therapeutic options worthy of exploring further in these challenging tumors. Moreover, our data showed that cells with an APC mutation are highly sensitive to TDP1 inhibitors. The identification of mutation(s) in the DNA repair genes such as in the APC gene by sequencing GEP NEC patient tumors could serve a biomarker for predicting vulnerability to TDP1 inhibitors.