IDH mutations are also frequently observed in other tumour types, such as acute myeloid leukaemia (AML), glioma, and cholangiocarcinoma [14]. Interestingly, the most common variant differs between the above-stated tumour types. Cartilage tumours and cholangiocarcinoma mainly have IDH1 p.R132C variants (~60%), glioma predominantly harbours IDH1 p.R132H mutations (~90%), and AML often has IDH2 p.R140Q mutations (~40%) [15,16]. None of the variants are exclusively observed in one tumour type, suggesting that different point mutations can have a similar effect on tumourigenesis, although the level of the oncometabolite D-2-hydroxyglutarate (D-2-HG) produced by these variants differs [17,18,19]. The prognostic value of IDH mutations in these tumour types is also diverse, and only glioma patients have a clear favourable outcome when their tumour harbours an IDH mutation [20,21,22,23]. Studies that were performed to determine the prognostic value of IDH mutations in chondrosarcoma show contradictory results. While it was previously reported that IDH mutations do not predict outcomes [15], other studies showed either a worse [24] or better [25] prognosis for IDH mutant (IDH^MUT) chondrosarcoma patients. The three patient cohorts were similar in size (n = 70 to 80) and median age (50 to 60 years), but the chondrosarcoma subtype inclusion (conventional versus addition of dedifferentiated and mesenchymal cases) and median follow-up time (4.3 versus ≥10 years) differed, which might explain the discrepancy in results. Another factor might be the type of technique used to assign patients to the IDH^MUT subgroup. For instance, Sanger sequencing is not sensitive enough to detect mutations when present in less than <30% of the sequenced PCR product, leading to false-negative results in samples with a low IDH^MUT variant allele frequency or tumour cell percentage and thereby the assignment of IDH^MUT patients to the IDH wildtype (IDH^WT) subgroup. Despite the lack of prognostic value, the high occurrence rate of IDH mutations in all of these tumour types suggests that they have an important role in driving tumourigenesis, already in the early stages of tumour development.

Both IDH enzymes function in the tricarboxylic acid (TCA) cycle, where they convert isocitrate into α-ketoglutarate (α-KG) and CO₂. Mutated IDH enzymes acquire a neomorphic function, leading to the additional conversion of α-KG into the oncometabolite D-2-HG [39]. The IDH1 p.R132C variant is one of the most efficient D-2-HG producers, while both IDH1 p.R132H and IDH2 p.R140Q produce lower levels of the oncometabolite [17,18,19]. As certain variants are more frequently observed in specific tumour types [15,16], this could suggest that chondrosarcoma and cholangiocarcinoma rely on high D-2-HG levels, while glioma and AML depend on relatively lower levels of the oncometabolite.

Due to the high structural similarity between α-KG and its antagonist D-2-HG, the oncometabolite is able to competitively bind α-KG-dependent enzymes, leading to the overall inhibition of this class of enzymes [40,41]. The inhibition of α-KG-dependent enzymes leads to widespread changes in the epigenomes and metabolomes of cells and affects DNA repair and cellular growth signalling pathways [42,43]. For instance, the D-2-HG-mediated inhibition of α-KG-dependent DNA demethylases (family of TET enzymes, including TET1/2) and histone demethylases (family of Jumonji enzymes, including KDMA4A/B) leads to an overall DNA hypermethylation phenotype, as well as an aberrant histone methylation phenotype in IDH mutant tumours. IDH^MUT enchondromas and chondrosarcomas are indeed characterised by a CpG island methylator phenotype (CIMP)-positive status, and DNA hypermethylation is present in primary IDH^MUT chondrosarcomas [7,44,45]. The family of Jumonji enzymes is also involved in the regulation of the Mechanistic Target Of Rapamycin Kinase (mTOR) signalling pathway, as well as DNA repair via the homologous recombination pathway. Moreover, IDH^MUT enzymes have a reduced ability to produce NADPH and consume high levels of NADPH to produce D-2-HG, resulting in severely reduced overall NADPH levels. This deficiency does not only cause metabolic stress but will also lead to an increase in reactive oxygen species (ROS), making IDH^MUT tumours more vulnerable to DNA damage. Besides the induction of metabolic stress, IDH^MUT tumours also undergo metabolic rewiring, including alterations in metabolites of the TCA cycle, a reduced dependency on glycolysis, and alterations in lipid metabolism. Additionally, D-2-HG-mediated inhibition of the prolyl hydroxylase domain proteins (EGLN1 and -2) leads to the upregulation of hypoxia-inducible factors (e.g., HIF1α), resulting in a metabolic switch to maintain oxygen homeostasis. D-2-HG also affects collagen maturation via the inhibition of proline and lysine hydroxylases (P4HA1-3 and PLOD1-3), leading to an impaired extracellular matrix structure. Thus, IDH mutations have a wide variety of downstream biological effects; therefore, these mutations are considered as the drivers in multiple tumour types.

To counteract the oncogenic activity of the IDH mutations, several inhibitors targeting either IDH1 p.R132 variants (e.g., ivosidenib) or IDH2 p.R140 variants (e.g., enasidenib) have been developed over the past couple of years [46]. In vitro studies and clinical trials show that AML patients could benefit from IDH^MUT protein inhibitors [26,32], although some patients develop resistance against these inhibitors over time. This acquired resistance is multi-factorial and can be caused by second-site mutations in IDH^MUT genes to prevent the binding of IDH^MUT protein inhibitors, IDH^MUT isoform switching to circumvent the effect of IDH^MUT protein inhibitors, or novel acquired mutations in genes encoding for receptor tyrosine kinases (RTKs) [33,34,35]. Direct inhibition of IDH^MUT proteins seems less promising for other tumour types that frequently harbour an IDH mutation [27,28,36,37]. Especially in chondrosarcoma, the effect of IDH^MUT protein inhibitors in in vitro assays seems controversial. While several studies have shown that IDH1^MUT protein inhibition does not affect the tumourigenic properties of chondrosarcoma cell lines [27,29], other groups have shown that IDH1^MUT protein inhibition causes a decreased proliferation rate in chondrosarcoma cell lines at higher doses or with a different compound [30,31]. Recent results from a phase I clinical trial with the IDH1^MUT inhibitor ivosidenib showed that prolonged disease control (i.e., progression-free survival of ~6 months) could be achieved in a subset of patients with advanced chondrosarcoma, predominantly in patients with a minimal number of co-occurring mutations [38].

As IDH^MUT protein inhibitors showed limited efficacy in in vitro assays and clinical trials or acquired resistance was observed, a large number of in vitro studies were performed to determine whether directly targeting the downstream biological effects of IDH mutations would be more promising. Indeed, multiple synthetic lethal interactions with the IDH mutation were reported for AML and glioma, including radiotherapy, chemotherapy, and agents that target poly(ADP-ribose) polymerase (PARP), B-cell lymphoma 2 (Bcl-2) family members, Bromodomain and Extra-Terminal Motif (BET) proteins, DNA methyltransferases (DNMTs), mTOR, Nicotinamide Phosphoribosyltransferase (NAMPT), and glutaminase [27,28,47,48,49,50,51,52,53,54,55,56,57,58,59,60]. However, chondrosarcoma cell lines are variably sensitive to a selection of these therapies, but the effect seems irrespective of the IDH mutation status, as IDH^WT chondrosarcoma cell lines show similar treatment responses [61,62,63,64,65,66,67].

These contradictory findings on synthetic lethal interactions with the IDH mutation might be ascribed to different factors. First, the cell of origin and the tumour microenvironment (e.g., cartilaginous matrix formation and hypoxia in chondrosarcoma) of the distinct tumour types that frequently harbour an IDH mutation are highly different and could therefore influence the role that IDH mutations play in tumourigenesis. Second, the level of the D-2-HG oncometabolite may also influence the downstream biological effects of IDH mutations. The most common IDH variants in AML and glioma both produce relatively low D-2-HG levels, whilst the most common point mutation in both cholangiocarcinoma and chondrosarcoma produces relatively high levels of the oncometabolite [17,18,19]. It was recently shown that a lower level of DNA hypermethylation was observed for the IDH1 p.R132H variant compared to non-p.R132H variants, irrespective of tumour type [16]. Lastly, the type of in vitro model (endogenous vs. artificially created) might influence whether synthetic lethal interactions with the IDH mutation are present or not. The introduction of an IDH mutation in a glioma model leads to reduced glutamine and glutamate levels, but this change in TCA cycle metabolites is not present when endogenous IDH^WT and IDH^MUT glioma models are compared [68]. Most synthetic lethal interactions with the IDH mutation were indeed identified in generic cancer cell lines with an introduced IDH^MUT. AML and glioma cell lines with an endogenous IDH^MUT are scarce, but the utilised chondrosarcoma cell lines do harbour endogenous IDH mutations and this difference in model type could explain why synthetic lethal interactions with the IDH mutation are absent in the chondrosarcoma in vitro studies. As IDH mutations occur early during tumourigenesis, especially in chondrosarcoma, artificial models with an introduced IDH mutation may not be representative of the role that IDH mutations normally play in tumourigenesis. These studies also introduced the IDH mutation in generic cancer cell lines that are easy to transfect (e.g., HeLa, HCT116, and U2OS cells), and these cell lines do not represent the tumour types in which IDH mutations frequently occur. Moreover, most studies generated models that overexpressed the IDH^MUT protein, whilst the balanced expression of IDH^WT and IDH^MUT is needed to retain efficient D-2-HG production [69].

In addition to these factors, it was recently shown that the (epi)genetic landscape in which IDH^MUT and IDH^WT are embedded is another important aspect to take into consideration when defining underlying vulnerabilities in tumour types that frequently harbour an IDH mutation. Studies on AML and glioma have shown that the genetic and epigenetic landscape in which IDH^WT and IDH^MUT function is highly heterogenous and thereby influences the therapy response and patient outcome [70,71,72,73,74,75,76,77,78,79,80,81]. For instance, mutations in TP53 and ATRX are the underlying denominator in defining which IDH^WT and IDH^MUT gliomas respond to radiotherapy [70]; the overexpression of BCAT1 in IDH^WT AML leads to an IDH^MUT-like DNA hypermethylation phenotype [71], and additional mutations in DNMT3A cause reduced levels of DNA hypermethylation in IDH^MUT AML samples [74]. Furthermore, co-occurring (epi)genetic alterations such as CIMP status [78], 1p19q deletions [80], CDKN2A deletions [78,79], MET amplifications [78], PDGFRA amplifications [79], and TERT mutations [80] influence overall survival in IDH^MUT glioma patients. Moreover, IDH^MUT AML patients with a co-occurring NPM1 mutation show overall a better response to chemotherapy with or without venetoclax [81]. The influence of co-occurring (epi)genetic alterations may also explain why distinct IDH^MUT tumour types differ in therapy sensitivity and underlines the need to use endogenous IDH^MUT models, as generic cancer cell lines with an introduced IDH mutation do not represent the (epi)genetic landscape in which IDH mutations naturally exist. Thus, the IDH mutation status does not solely define the underlying vulnerabilities, which is in line with previous findings for chondrosarcoma [61,62,63,64,65,66,67], suggesting that a dichotomy between IDH^WT and IDH^MUT is too simplistic.

Besides IDH mutations, chondrosarcomas frequently harbour mutations in TP53, CDKN2A/B, COL2A1, YEATS2, NRAS, and TERT [82,83,84,85,86]. However, the rest of the previously observed co-occurring mutations seem to follow a more random pattern and are present in less than 10% of the chondrosarcomas [25,83,84,87], leading to a highly heterogeneous genetic landscape in which IDH^WT and IDH^MUT function in chondrosarcoma. Furthermore, IDH^MUT chondrosarcomas are characterised by a global hypermethylation phenotype that changes with increasing histological grade [44,45], and, based on methylation profiles alone, several chondrosarcoma subgroups could be defined, even within IDH^WT and IDH^MUT tumours [88]. Moreover, using chondrosarcoma transcriptome and methylome data, it was previously shown that different molecular subtypes (i.e., high mitotic state, 14q32 miRNA cluster loss of expression, and IDH^MUT-induced DNA hypermethylation) exist, and that these are associated with patient outcomes [89]. Moreover, (epi)genetic alterations in the TERT gene (i.e., hypermethylation and promotor mutations) affect the survival probability of IDH1^MUT chondrosarcoma patients, whilst this association is absent in IDH^WT and IDH2^MUT patients [87].