Alkylating-induced hypermutation in pancreatic neuroendocrine tumours

Introduction Pancreatic neuroendocrine tumours (PanNET) represent 1%–2% of all pancreatic neoplasms and about 9% of all gastroenteropancreatic neuroendocrine neoplasms. 1 2 The incidence of PanNET has increased over recent decades, likely due to early detection, greater clinical awareness and a potential genuine increase.3 Metastatic PanNET portend a worse prognosis compared with small bowel NET primaries. This poorer prognosis is multifactorial, involving higher tumour grades and aggressiveness, delayed and often late-stage diagnosis due to lack of tumour-related symptoms in most cases, surgical challenges due to anatomical location and distinct genetic profiles.PanNET exhibit unique genetic profile characterised by frequent mutations in MEN1 (histone modification), DAXX and ATRX (chromatin remodelling and telomere maintenance) and in TSC2, PTEN and PIK3CA (part of the mTOR signalling pathway).4 5 Conversely, small bowel NETs frequently display genetic alterations in CDKN1B (Wnt/β-catenin pathway).6 Overall, neuroendocrine neoplasms harbour low tumour mutational burden (TMB), with 0.1 to 5 mutations per megabase (mut/Mb) in PanNET and small bowel NET and 5–10 mut/Mb in microsatellite stable extra-pulmonary neuroendocrine carcinomas (NEC).7–10 However, the phenomenon of high TMB has been described in PanNET following treatment with alkylating agents.Chemotherapy is an important therapeutic modality for patients with PanNET. Despite recent advances in available treatments, including data on the use of peptide receptor radionuclide therapy (PRRT) and cabozantinib, alkylating-based chemotherapy remains commonly used.11 12 Temozolomide (TMZ) has replaced early-generation alkylating agents such as streptozotocin and dacarbazine and can be used either alone or in combination with capecitabine (an oral analogue of 5-fluorouracil).13 For patients with grades 1–2 (G1–G2) advanced (unresectable or metastatic) PanNET, TMZ combined with capecitabine (CAPTEM) versus TMZ alone led to a significant median progression-free survival gain of nearly 1 year.13 Based on these results, the European Society for Medical Oncology, the European Neuroendocrine Tumor Society (ENETS) and the American Society of Clinical Oncology list alkylating agents as a potential first-line option for the treatment of advanced PanNET with grade 2 and grade 3 (with a Ki-67 index less than 50%).14–16 However, despite the anti-tumour effects of TMZ in PanNET, there is emerging evidence that it can induce tumour grade progression and mutations.17In glioblastoma, TMZ induces a hypermutator phenotype in 15%–20% of tumours, primarily through loss-of-function mutations in MSH6 leading to mismatch repair (MMR) deficiency.18–21 In advanced colorectal cancer, the MAYA phase II trial investigated TMZ priming in patients with O6-methylguanine-DNA methyltransferase (MGMT) deficiency/MMR proficient to evaluate whether there was an increase in TMB after at least two cycles of TMZ and, if so, whether ipilimumab and nivolumab, both immune checkpoint inhibitors, would be effective in controlling tumour growth.22 23 In the MAYA trial, TMB of patients who achieved a progression-free survival of at least 8 months, the primary endpoint, increased from 1.8 to 3.4 mut/Mb at baseline to 11 to 161 mut/Mb.19 In all cases of TMB-high, loss of function mutations in MMR genes in post-alkylating biopsies were identified. Unexpectedly, despite harbouring TMB-high, colorectal cancers were not responsive to immune checkpoint inhibitors.19 In the glioblastoma case reports, benefits from immunotherapy were also not demonstrated.18–21 The phenomenon of TMB-high has also been described in PanNET, with much discussion about the efficacy of immune checkpoint inhibitors in this setting.This review provides an updated synthesis of data on the hypermutator phenotype in PanNET, focusing on clinical characteristics, prognosis and the role of immune checkpoint inhibitors. For this narrative review, we searched PubMed for articles published between 2015 and 2025 using terms including PanNET, TMZ, alkylating agents, tumour mutational burden, hypermutation and immune checkpoint inhibitor. Additional references were identified by screening bibliographies of relevant studies presented at the ENETS, the North American Neuroendocrine Tumor Society and at the European Society of Medical Oncology congresses.Grade progression following alkylating agents Metastatic PanNET may progress from a low/intermediate to a high-grade (G3) disease in approximately 30%–40% of cases. 24–27 Although the molecular mechanisms underlying the acquired aggressive traits are unknown, evidence is accumulating regarding the role of therapeutically induced mutagenesis, especially alkylating agents.TMZ is the most widely used alkylating agent in the treatment of PanNET. It is an oral prodrug that undergoes spontaneous hydrolysis to generate its active metabolite, which methylates DNA primarily at the O6 position of guanine. This modification results in mispairing with thymine during DNA replication, thereby inducing genomic instability, apoptosis and subsequent cell death. However, the cytotoxic effects of TMZ can be counteracted by the DNA repair enzyme MGMT, which removes the methyl group from O6-methylguanine. Consequently, decreased MGMT activity enhances sensitivity to alkylating agents by impairing DNA repair.28Alkylating exposure is a potential contributor to grade progression in PanNET, with grade progression being reported in 40% to 60% of PanNET, depending on the timing of biopsy after treatment. In a retrospective monocentric cohort, by Mestier et al, patients with metastatic MGMT-deficient G1/G2 PanNET treated with TMZ, with or without capecitabine, reportedly experienced histological progression to G3 disease in 50% of cases, with the median Ki-67 index increasing from 6% prior to treatment to 40% following TMZ exposure. This grade progression was associated with poorer clinical outcomes, including a shorter time to TMZ failure and reduced overall survival from treatment initiation.29 Another retrospective study, by Trevisani et al, analysed 53 patients with advanced PanNET, comprising seven G1, 22 G2 and 23 G3 tumours at diagnosis. Similarly, 43% (six out of 14 patients with data available) experienced an increase in tumour grade from G1/G2 to G3 following TMZ exposure.30 Although potentially overestimated due to patient selection bias, a validation cohort in the study by Backman et al reported G3 progression in 67% (16 of 24) PanNET following treatment with alkylating agents. Notably, half of the G3 PanNET identified that post-therapy exhibited a Ki-67 proliferation index exceeding 50%.31 32Prior exposure to radioligand therapy may act as a confounding or contributing factor to grade progression in PanNET. In a study by Cordero-Hernandez et al, among 39 patients with G1/G2 PanNET who received at least one cycle of PRRT, seven cases exhibited high-grade progression; all had previously been treated with alkylating agents.33Alkylating drugs and hypermutation TMZ, according to recent and accumulating evidence, may induce genetic alterations and contribute to the development of a hypermutated phenotype during the treatment of patients with PanNET. In the study by Mestier et al, molecular analyses were performed in 25 patients for whom paired pre- and post-TMZ tumour samples were available. Post-treatment, there was an increased frequency of alterations in genes involved in cell cycle regulation, as well as the MAPK and mTOR signalling pathways. A hypermutator phenotype, defined as a TMB exceeding 30 mut/Mb, was identified in 26% (6/26) evaluable patients, five of whom had progressed to G3 disease.25 An independent validation cohort of 1079 PanNET, genetically profiled using the Foundation Medicine panel, identified 25 cases (2.3%) with high TMB. Among these, 67% exhibited an alkylating agent-associated mutational signature, although treatment data were not available.29 Yet, both studies performed tumour biopsies at different time points, not necessarily on disease progression.A multicentre retrospective study was performed on Italian and Brazilian patients with PanNET who had tumour samples collected prior to and on progression on alkylating drugs.34 The group performed next-generation sequencing (NGS) focused on alterations in DNA damage repair (DDR) pathways, assessing genes involved in homologous recombination (ATM, ATR, CHEK2, BRCA1/2, RAD51, PALB2), base excision repair (MUTYH, POLE, POLQ) and MMR pathways (MSH2, MSH6, MLH1, PMS2). Among 23 pre- and post-TMZ PanNET samples with available NGS data, TMB-high tumours had more DDR alterations (60% vs 18% of TMB-low; p=0.05) and mutations in MMR genes (60% vs 0%; p=0.0004). Patients with longer TMZ exposure (median of 10 and six cycles) and those also pretreated with radioligand therapy (60% vs 22.7%; p=0.03) had a higher frequency of TMB-high PanNET. As expected, tumours with high TMB were more likely to develop grade progression: 85% of TMB-high cases presented a conversion to G3 in comparison to 20% of TMB-low PanNET.The Italian group updated their results in a larger sample of pre-TMZ cases with available NGS data. Prior to TMZ, DDR alterations were identified in 15% (7/45) of cases, of which 70% were clonal (defined by a variant allele frequency (>10%)). Post-TMZ analysis revealed DDR alterations in 36% (5/14) of cases, of which only 7% were clonal. TMB-high, defined as >15 mut/Mb, was observed in five of 14 (36%) post-treatment samples, with 7% being clonal mutations. Among six patients with matched pre- and post-TMZ tumour samples, three (50%) acquired a hypermutated phenotype following treatment.30 These data suggest that alkylating agents, with or without radioligand therapy, contribute to the emergence of a hypermutator phenotype in approximately 30% of PanNET through both direct mutagenesis and selective clonal pressure. Specifically, the alkylating-induced hypermutation is frequently accompanied by a predominance of single base substitution signature 11 (SBS11), with C>T transitions.Tumour-related factors may also contribute to the onset of hypermutation. A recent multicenter study reported that pathogenic variants in MUYTH in patients with PanNET are associated with aggressive disease course.35 Of seven tumours molecularly profiled after alkylating drugs and/or radioligand therapy, three developed TMB-high.Checkpoint inhibitor efficacy in hypermutated PanNET Case reports have shown impressive sustained partial responses in patients with alkylating-induced TMB-high PanNET treated with immune checkpoint inhibitors. 36–38 One patient with an MSI-high tumour and 2134 mut/Mb remains disease-free after 2 years in nivolumab and ipilimumab.38 More recently, a multinational retrospective study was conducted to evaluate the efficacy of immune checkpoint inhibitors in patients with advanced PanNET.39 40 The cohort included 53 patients treated with alkylating-based regimens (mostly TMZ) regardless of TMB-high, with tumours predominantly classified as G2 (40%) and a pretreatment median Ki-67 proliferation index of 21%. Nearly half of patients also received radioligand therapy. Immune checkpoint inhibitors were primarily administered as a combination of anti-PD-1/PD-L1 and anti-CTLA-4 agents (70%). The observed overall response rate was 19%, the disease control rate was 49% and the median progression-free survival was only 3.3 months. Among the 38 patients with tumour tissue or circulating tumour DNA evaluated by NGS, 55% (29/53) exhibited TMB-high post an alkylating drug, defined as TMB greater than 10 mut/Mb. Additionally, 14 of 35 patients had tumours assessed for MMR status, using either mutational analysis or immunohistochemistry, and showed evidence of MMR alterations. Patients with TMB-high demonstrated significantly higher objective response (35% vs 0%, p=0.01) and longer median progression-free survival (4.9 vs 2.8 months, p=0.019) compared with those with low or unknown TMB. Interestingly, patients with TMB-high and somatic MMR alterations experienced further improved outcomes, with an objective response of 50% versus 8% (p=0.002) and a median progression-free survival of 12.1 versus 2.8 months (p=0.006), respectively. In this series, there were no differences in response rates between anti-PD-1 monotherapy antibodies or combined with an anti-CTL-4 antibody.The reasons why some patients with TMB-high and MMR defects still do not derive benefit from immune checkpoint inhibitors remain obscure. One hypothesis is heterogeneity across metastases. Mestier et al reported two illustrative cases of patients with PanNET who developed TMB-high following CAPTEM and subsequently received pembrolizumab. Both cases exhibited discordant responses to immune checkpoint inhibitors, with regression of some lesions and progression of others. Intertumoral heterogeneity might drive the discordant efficacy of immunotherapy, with different lesions showing different TMB (4–322) and MMR alterations.29 Moreover, although pre- and post-TMZ samples reveal a significant rise in alterations of DNA DDR and MMR genes, many of these mutations are subclonal and do not result in microsatellite instability.41 This observation suggests possible functional suppression of MMR genes without complete pathway inactivation, potentially limiting the efficacy of immunotherapy in this context. Beyond tumour heterogeneity and subclonal MMR mutations, other mechanisms such as loss of antigen presentation (eg, B2M mutations), alternative checkpoint upregulation or an immunosuppressive microenvironment may also contribute to immunotherapy resistance in hypermutated PanNET and deserve investigation.42 43Clinical implications There are important clinical implications about the onset of the TMB-high phenomenon in PanNET. First, there is a clinical perception of high-grade transformation, with aggressive evolution, rapid increase of number and size of metastases. Three patients from our collaborative study 31 (unpublished data) had very aggressive tumour progression after TMZ, with patients dying in a few weeks, precluding somatic NGS evaluation due to poor clinical conditions. Second, definitions of TMB-high in PanNET vary across studies (10–30 mut/Mb) influenced by differences in sequencing platforms and panel sizes; nonetheless, once TMB-high is detected, an immune checkpoint inhibitor can be considered, based on the KEYNOTE-158 clinical trial.44 Third, if an immune checkpoint inhibitor is not available, platinum-based chemotherapy such as cisplatin combined with etoposide or irinotecan, which are recommended for NEC, can be used. FOLFIRINOX (5-fluorouracil, irinotecan and oxaliplatin) is also an option for aggressive PanNET.45 46We also think the evaluation of the therapeutical sequencing of patients with PanNET in future studies is essential. Since the principle of ‘first do not harm’ is the basis of any medical intervention, we propose the treatment sequencing of patients with PanNET should be done with caution so as not to put excessive ‘treatment pressure’ on tumours. Until we understand the mechanisms of TMB-high transformation and provide effective treatments for these patients, it is advisable to sequence treatments for PanNET patients in a ‘gentler’ fashion. Therefore, we propose the following strategies which are summarised in table 1 that (1) alkylating agents could be stopped after 12–18 cycles instead of administering treatment until progression; (2) maintenance therapy with somatostatin analogues (for tumours which remain somatostatin-receptor positive after this period), capecitabine (for somatostatin-receptor negative tumours) or even chemoholiday (in low grade tumours) can be considered; (3) for progressive PanNET post-alkylating drugs, avoid immediate sequential therapy of alkylating and radioligand therapy or vice versa; instead, bridge treatments with either everolimus, a tyrosine kinase inhibitor or liver-directed therapies, if possible; (4) if available, perform NGS in a new tumour tissue obtained by biopsy or by ctDNA after a few cycles of alkylating agents to detect TMB-high early instead of only on aggressive tumour progression; (5) monitor closely for signs and symptoms of aggressive tumour progression such as new symptoms of pain, weight loss and palpable masses, with radiological exams every 2 to 3 months of alkylating drugs or radioligand therapy and with monthly consultations.Table 1Proposed strategies to manage treatment-induced hypermutation of pancreatic neuroendocrine tumoursAimRationalProposed strategyMinimise the risk of hypermutationLimit the cumulative dose of alkylating agentStop alkylating agents after 12 cycles, favouring maintenance therapy until progressionReduce synergistic risk of hypermutationBoth alkylating agents and radioligand therapy are risk factors for hypermutation in pancreatic neuroendocrine tumoursAvoid immediate sequential therapy of alkylating and radioligand therapy or vice versa; instead, bridge with other treatmentsEarly detection of hypermutationHypermutated pancreatic neuroendocrine tumours tend to be very aggressive.If available, perform tumour NGS after a few cycles of alkylating agents to detect hypermutation earlyEarly diagnosis of aggressive evolutionHypermutated pancreatic neuroendocrine tumours tend to be very aggressive.Monitor closely for signs and symptoms of aggressive tumour progression with radiological exams every 2 to 3 months of alkylating drugs or radioligand therapy and with monthly consultationsNGS, next-generation sequencing.Noteworthily, a key limitation is that our conclusions are primarily drawn from retrospective studies and small cohorts, with the majority of data derived from conference abstracts. Yet, these findings are consistent across multiple groups. Once full peer-reviewed data are available, these observations should be validated in larger studies.Future perspective While TMZ increases the risk of hypermutation in PanNET, it remains unclear whether this phenomenon is exclusive to alkylating agents or is a broader feature of DNA-damaging therapies. Platinum-based chemotherapy, such as oxaliplatin, also induces DNA crosslinking and replication stress, suggesting a potential for similar mutagenic effects. Likewise, currently approved radioligand therapy (Lutetium 177 DOTATATE) causes double-stranded DNA breaks via β-emission and may lead to cumulative mutagenic effects. The extent to which radioligand therapy and alkylating drugs potentiate each other to induce tumour hypermutation remains to be determined by larger studies. Nonetheless, it may become more common with the sequential administration of multiple DNA-damaging radiopharmaceuticals under development. Somatostatin receptors compounded with alpha emitters, such as Actinium225 and Plumb212, are being investigated in patients with PanNET after progression on the β-emitter Lutetium177 DOTATATE.47 Because many of these patients will also be treated with TMZ-based regimens, the incidence of TMB-high is likely to increase due to cumulative toxic effects on cancer cells’ DNA. Also, both alkylating agents and radioligand therapies are associated with clonal haematopoiesis and potentially, myelodysplastic syndrome and secondary leukaemia.48Current data report variable outcomes from immune checkpoint inhibitors. Dual checkpoint blockade (eg, anti-PD-1/PD-L1 with anti-CTLA-4) seems promising, even in the setting of prior progression on anti-PD-1 monotherapy.34 However, we still do not know what immunotherapy strategy is the best in TMB-h PanNET. Further treatment optimisation may involve integrating agents that enhance tumour antigenicity or immune priming. It is also important to understand the reasons underlying the lack of efficacy of immune checkpoint inhibitors in this setting. We are currently evaluating the immune profiles of tumour-infiltrating cells in samples of patients pretreated with TMZ to identify potential predictive factors to immunotherapy response.Finally, an emerging avenue for optimising treatment strategies is the evaluation of molecular biomarkers using liquid biopsies. ctDNA has shown potential for capturing tumour heterogeneity and tracking clonal evolution in response to therapy. Future research should investigate the kinetics of ctDNA analysis to enable real-time monitoring of TMB and early detection of TMB-high, MMR deficiency and DDR alterations. However, discrepancies between tissue and ctDNA findings may occur, underlining the importance of harmonising sampling approaches and validating thresholds to define TMB-high or MMR-deficient disease in liquid biopsies.In conclusion, despite the small number of patients and the retrospective nature of studies evaluating TMB-high PanNET, this phenomenon seems to arise in approximately one-third of patients pretreated with alkylating agents and is associated with an aggressive course (figure 1). Immune checkpoint inhibitors are effective in a subset of patients, including cases with complete and durable responses. The co-occurrence of TMB-high and MMR alterations increases the odds of tumour response to immunotherapy. Therefore, if available, we recommend TMB should be assessed in patients with tumour progression on alkylating agents. Importantly, treatment sequencing with alkylating agents and radioligand therapies may increase the incidence of TMB-high PanNET, supporting close monitoring of these patients for signs of tumour transformation.Figure 1Alkylating-induced hypermutation in PanNET and its clinical implications. Schematic representation of the development of hypermutation phenotype in PanNET following exposure to alkylating agents with or without PRRT. Treatment exposure may lead to increased tumour grade and Ki 67, TMB-high, DDR/MMR mutations (such as in MSH6), presence of mutational signature SBS11, and aggressive progression. Post-progression treatment options include immune checkpoint inhibitors or clinical trials. DDR, DNA damage repair; MMR, mismatch repair; PanNET, pancreatic neuroendocrine tumours; PRRT, peptide receptor radionuclide therapy.