Comprehensive Immunogenomic Profiling of IDH1-/2-Altered Cholangiocarcinoma

Abstract

PURPOSE

Isocitrate dehydrogenase (IDH)1/2 genomic alterations (GA) occur in 20% of intrahepatic cholangiocarcinoma (iCCA); however, the immunogenomic landscape of IDH1-/2-mutated iCCA is largely unknown.

METHODS

Comprehensive genomic profiling (CGP) was performed on 3,067 cases of advanced iCCA. Tumor mutational burden (TMB), PD-L1 expression (Dako 22C3), microsatellite instability (MSI), and genomic loss of heterozygosity (gLOH) as a surrogate marker for homologous recombination deficiency were examined. RNA sequencing of 73 patient samples was analyzed for differences in stromal/immune cell infiltration, immune marker expression, and T-cell inflammation. Tissue microarray arrays were subjected to multiplex immunohistochemistry and colocalization analysis in 100 surgical samples. Retrospective clinical data were collected for 501 patients with cholangiocarcinoma to examine median overall survival (mOS) in IDH1/2+ versus IDHwt.

RESULTS

Of 3,067 iCCA cases subjected to CGP, 426 (14%) were IDH1+ and 125 (4%) were IDH2+. IDH1 GA included R132C (69%) and R132L/G/S/H/F (16%/7%/4%/3%/<1%). IDH2 GA occurred at R172 (94.4%) and R140 (6.6%). No significant difference was seen in median gLOH between IDH1+ versus IDHwt iCCA (P = .37), although patterns of comutations differed. MSI-High (P = .009), TMB ≥10 mut/Mb (P < .0001), and PD-L1 positivity were lower in IDH1/2+ versus IDHwt iCCA. Resting natural killer cell population, CD70, and programmed cell death 1 expression were significantly higher in non–IDH1-mutated cases, whereas V-set domain containing T-cell activation inhibitor 1 (B7-H4) expression was significantly higher in IDH1+. No significant difference in mOS was observed between IDH1/2+ versus IDHwt patients.

CONCLUSION

Significant differences in GA and immune biomarkers are noted between IDH1/2+ and IDHwt iCCA. IDH1-/2-mutated tumors appear immunologically cold without gLOH. These immunogenomic data provide insight for precision targeting of iCCA with IDH alterations.

Comprehensive genomic profiling and immune biomarker analysis in IDH1-/2-altered cholangiocarcinoma.

INTRODUCTION

Intrahepatic cholangiocarcinoma (iCCA) is a molecularly diverse malignancy arising from the second-order bile ducts within the liver.¹ In the past decade, significant advances in the molecular profiling of cholangiocarcinomas (CCAs) have resulted in identification of targetable molecular alterations and enriched signaling pathways.^2-8 Fibroblast growth factor receptor 2 (FGFR2) fusions, and isocitrate dehydrogenase (IDH1/2) and BAP1 mutations are frequently altered in iCCA, while KRAS, TP53, and SMAD4 alterations are more frequently observed in extrahepatic CCA. Activating alterations in receptor protein-tyrosine kinase ERBB2/ERBB3 have been frequently found in gallbladder carcinoma.^9,10 This has led to the recent approval of several small molecule inhibitors targeting the FGFR and IDH1 pathways in CCA, and many others in the pipeline.^11-14

CONTEXT

• Key Objective

• Intrahepatic cholangiocarcinoma is a molecularly diverse malignancy with multiple targetable genomic alterations including mutations in isocitrate dehydrogenase (IDH) 1/2. This analysis examined two data sets of patients (n = 3,067 and n = 501) and, to our knowledge, provides the largest known descriptive analysis of the genomic and immune microenvironment of IDH1-/2-altered versus IDH wildtype (WT) cholangiocarcinoma.

• Knowledge Generated

• Significant differences in the co-occurring mutational profiles and biomarkers of immunotherapy response were noted between IDH1-/2-altered and IDHwt cholangiocarcinoma, with IDH-mutated tumors emerging as more immunologically cold. Preliminary analysis showed immune biomarker differences between the two subgroups.

• Relevance

• With the recent approval of the IDH1 inhibitor ivosidenib, and other IDH inhibitors in the pipeline, the findings from this study will be informative in planning future precision immune-oncology therapies in cholangiocarcinoma.

IDH enzymes catalyze the oxidative decarboxylation of isocitrate to α-ketoglutarate (αKG) while reducing NADP⁺ to NADPH.¹⁵ Mutated IDH enzymes result in the abnormal production of the oncometabolite 2-hydroxyglutarate (2-HG), which plays a role in tumorigenesis by competitively inhibiting other cellular enzymes that regulate DNA repair, cellular metabolism, and epigenetics.¹⁵ IDH1/2 mutations are seen in 13%-15% of patients with iCCA, compared with <1% in patients with extrahepatic CCA.¹⁶

The small-molecule IDH1 inhibitor ivosidenib was studied in the ClarIDHy trial, a phase III randomized, double-blind, placebo-controlled study of previously treated patients with IDH1-mutated CCA.¹⁴ Ivosidenib showed a modest improvement in median progression-free survival (2.7 months) compared with placebo (1.4 months; hazard ratio [HR], 0.37; P < .0001) and median overall survival (mOS) of 10.3 months versus 7.5 months (HR, 0.79, P = .09).^14,17 Ivosidenib received (US) Food and Drug Administration (FDA) approval for previously treated patients with advanced or metastatic CCA with an IDH1 mutation in August 2021. Several other IDH1 inhibitors are under study alone or in combination.

IDH-altered CCA has also been linked to increased DNA damage. 2-HG has been shown to inhibit homologous recombination repair and impair DNA damage repair.¹⁸ Gliomas containing IDH mutations have improved outcomes to radiation therapy, and poly(adenosine 5′-diphosphate) ribose polymerase inhibitor (PARP) inhibitors are being explored in IDH-mutated solid malignancies.^19,20

Previous comprehensive molecular profiling studies have examined the molecular alterations in patients with biliary tract cancers; however, the molecular landscape and immune microenvironment of IDH-altered CCA are relatively unexplored. In this study, we evaluated multiple data sets at the molecular, protein, and patient outcome levels to study differences with IDH1-/2-altered and IDHwt CCA.

METHODS

Comprehensive Genomic Profiling

In total 3,067 cases of advanced iCCA were subjected to comprehensive genomic profiling (CGP). A hybrid capture–based next‐generation sequencing assay evaluating base substitutions, short insertion and deletion alterations (INDELs), rearrangements/fusions, and copy-number alterations in 324 genes and select gene rearrangements was performed.²¹ This study was conducted in accordance with the institutional review board (IRB; protocol PA13-0206). In addition, approval for this study, including a waiver of informed consent and a HIPAA waiver of authorization, was obtained from the Western IRB (Protocol No. 20152817).

PD-L1, MSI, TMB, and LOH Testing

PD-L1 expression in tumor cells (tumor proportion score [TPS]) was measured by immunohistochemistry (IHC; Dako 22C3). High positivity was defined as ≥50% expression and low positivity was defined as ≥1% and <50% expression. Microsatellite instability (MSI) was determined on 95-114 loci with exclusion of cases in which MSI could not be determined in MSI-high frequency calculation.²² Tumor mutation burden (TMB) was determined on 0.8-1.11 Mb of sequenced DNA.²³ Loss of heterozygosity (LOH) was determined by leveraging estimated copy number and minor allele count of single nucleotide polymorphisms across interrogated genome and is reported as a percentage.^24,25

Immunohistochemical Whole Slide and Image Analysis

One hundred formalin-fixed paraffin-embedded surgical samples from 96 patients treated at MD Anderson Cancer Center between 2004 and 2016 were subjected to IHC using 14 antibodies (CD1, CD4, CD8, CD68, PD-1, PD-L1, B7-H3, B7-H4, IDO-1, VISTA, ICOS, OX40, TIM-3, and LAG-2) as previously described.²⁶

Tissue Microarray and Multiplex Immunofluorescence Analysis

Two paraffin block tissue microarrays (TMA) were constructed using two representative tumor samples (1-mm diameter cores) from 83 patient samples. TMA slides sectioned at 5 µm were subjected to multiplex immunofluorescence (mIF) as previously described,²⁶ using 12 antibodies multiplexed in two panels of six markers (Panel 1: CD3, CD8, CD58, PD1, PD-L1, and CK [pan-cytokeratin to identify tumor cells], and Panel 2: CD3, CD8, CD45-RO, FOXP-3, Granzyme B, and CK). A Vectra high-resolution imaging system was used for scanning and histologic analysis performed to quantify tumor purity. Multiplex stain individualization and unmixing as well as immune cell population identification and quantification was performed using InForm 2.2 image analysis software (PerkinElmer, Waltham, MA) and spectral library information used for fluorochrome and mIF marker association.²⁶

RNA Sequencing and Computation Analysis

RNA sequencing was performed using methods previously described.²⁷ Batch-corrected, transcripts per million–normalized transcriptomics data were log-transformed. Comparisons between IDH1-mutated and IDHwt samples were performed using the Estimation of STromal and Immune cells in MAlignant Tumours using Expression (ESTIMATE) data algorithm to assess stromal and immune cell infiltration.²⁸ ESTIMATE score was determined as the sum of the two on the basis of RNA expression data. Cell-type Identification by Estimating Relative Subsets of RNA Transcripts (CIBERSORT) algorithm was used to assess values of 22 immune cell types using mRNA expression data.²⁹ Calculations were performed in absolute mode. Immune checkpoint, epithelial to mesenchymal (EMT), vascularization, INF-gamma expression scores, and T-cell inflammation signature were measured with the Immuno-oncology gene interaction Maps (ImogiMap) package.^30-34 The two groups were compared using a Wilcoxon test with a Benjamini-Hochberg FDR correction applied.

Statistical Analyses

Statistical analyses for CGP were performed using R software (R Foundation for Statistical Computing, Vienna, Austria v3.6.0). Chi-square test or Fisher's exact test was used to compare proportions of categorical variables. Retrospective chart reviews were performed on a database of 501 patients with cholangiocarcinoma treated at MD Anderson Cancer Center diagnosed between 2008 and 2022. The Kaplan-Meier method was used to estimate probability of OS and the median OS, and log-rank test was used to compare OS between groups with and without IDH mutations. Statistical analyses were performed using R software (R Foundation for Statistical Computing, Vienna, Austria v3.6.0).

RESULTS

IDH1/2 Mutational Profile in iCCA

A total of 3,067 cases of advanced iCCA were subjected to CGP (Fig 1). IDH1- and IDH2-mutated (mut) iCCA were more often seen in females (P = .0002) and had a similar age distribution as IDH wildtype (WT) patients. Four-hundred and twenty-six of 3,067 (14%) patients had IDH1 genetic alterations and 125 (4%) had IDH2 alterations. IDH1 and IDH2 mutations were mutually exclusive. All (100%) IDH1 alterations were neomorphic and included R132C (69%), R132L (16%), R132G (8%), R132S (4%), R132H (3%), and R132F and 119Q (<1%). The distribution and frequency of IDH1/2 alterations is shown in Figure 2A. All genomic alterations (GA) in IDH1 and IDH2 were short variant base substitutions. No INDELs, rearrangements, or copy-number changes were identified.

FIG 1.

Flow diagram of study analyses with number of patients (N) noted. 3,067 iCCA cases from Foundation Medicine database and 501 patients with CCA treated at MD Anderson Cancer Center were included in the analyses. ^aOf the patient cohorts, 39 of 71 IDH1 cases from MDA had FM data and nine of 16 IDH2 cases from MDA had FM data. ^bDifferences in stromal and immune cell infiltration were evaluated between IDHmut and IDHwt cases with Estimation of STromal and Immune cells in MAlignant Tumours (ESTIMATE) algorithm, immune cell types were evaluated via Cell-type Identification by Estimating Relative Subsets of RNA Transcripts (CIBERSORT) algorithm, and immune cell signatures via ImogiMap algorithm. ^cFourteen IHC markers and two mIF panels evaluated for differences between IDHmut and IDHwt. FM, Foundation Medicine; MDA, MD Anderson Cancer Center; NGS, next-generation sequencing; iCCA, intrahepatic cholangiocarcinoma; IHC, immunohistochemistry; mIF, multiplex immunofluorescence.

FIG 2.

IDH1/2 mutational profile in iCCA. (A) Pie charts depict IDH1-/2-mutated cases within all iCCA (left) and the distribution of IDH1 (center) and IDH2 (right) mutations by frequency. (B) The top 25 most common co-occurring mutations depicted by IDH mutation status. iCCA, intrahepatic cholangiocarcinoma; IDH, isocitrate dehydrogenase.

Co-Occurring GA With IDH1/2 Mutations in iCCA

Frequency of co-occurring mutations in IDH1mut, IDH2mut, and IDHwt cases was assessed (Fig 2B). A range of 3.0-4.4 GA per tumor sample were noted in the patients with iCCA. ARID1A, BAP1, and PBRM1 alterations occurred most frequently in the IDH1- and IDH2-mutated patients compared with WT. Both IDH1mut and IDH2mut iCCA had significantly lower frequencies of other nontargetable GA, including TP53, CDKN2A/B, KRAS non G12C, MTAP, TERT, SMAD4, and MYC, compared with IDHwt patients (P values range from = 0.007 to <0.0001). IDH1mut and IDH2mut iCCA also had lower targetable GA than IDHwt patients, including FGFR2 rearrangements (P < .0001), ERBB2 (P = .0009), and BRAF (P = .04; Table 1). Co-occurring FGFR2 fusions were identified in two cases of IDH1mut (FGFR2-MYLK fusion and FGFR2-FHL2 fusion) and one case of IDH2mut (FGFR2-ATF fusion) iCCA. A list of co-occurring GA with IDH1/2 mutation status is summarized in Appendix Table A1.

TABLE 1.

Distribution of Common Co-Occurring, Other Targetable Mutations, and Markers of Immunotherapy Response Across IDH1-/2-Mutated and Wildtype CCA

Alteration/Marker	IDH1+ iCCA (n = 426)	IDH2+ iCCA (n = 125)	IDHwt iCCA (n = 2516)	P a
Most common co-occurring alterations, %
TP53	12	9	39	<.0001
CDKN2A/CDKN2B	20/16	12/9	33/23	<.0001
MTAP	8	6	17	<.0001
Co-occurring targetable alterations, %
FGFR2 rearrangements	1	1	10	<.0001
ERBB2 variants/amplifications	<1/2	2/1	1/4	.0009
BRAF	4	2	5	.04
Markers of immunotherapy response, %
MSI-high	<1	0	1	<.009
TMB ≥10 mut/Mb	<1	0	5	<.0001
PD-L1–low positive (TPS 1%-49%)	2 (n = 121)	17 (n = 35)	18 (n = 691)	<.001
PD-L1–high positive (TPS ≥50%)	1	9	6	.11

Abbreviations: CCA, cholangiocarcinoma; FGFR2, fibroblast growth factor receptor 2; iCCA, intrahepatic cholangiocarcinoma; MSI, microsatellite instability; TMB, tumor mutational burden; TPS, tumor proportion score.

^a P value between IDH1/2+ and IDHwt.

gLOH in IDH-Altered iCCA

Given the relationship between IDH mutations and homologous recombination deficiency (HRD), we examined genomic loss of heterozygosity (gLOH) in IDH1-/2-mutated and IDHwt patients, as well as BRCA-mutated iCCA as an additional comparator (Fig 3A). No significant difference was found between IDH1-/2-mutated and IDHwt patients (gLOH of 10.37%, 10.55%, and 10.21% in IDH1mut, IDH2mut, and IDHwt iCCA, respectively, P = .37).

FIG 3.

(A) Median gLOH, TMB, and PD-L1 expression between IDH1-/2-mutated and wildtype cases. As an additional comparator, gLOH in BRCA-mutated cohort shown with median gLOH of 16.36%. (B) IDHwt iCCA had significantly higher TMB than IDH1- and IDH2-mutated iCCA cases. (C) Representative slides of PD-L1–high positive (>50% TPS) in IDHwt (top panel) and negative (0% TPS) in IDH1 R132Cmut (lower panel). IDHwt iCCA had significantly higher low positive (TPS 1%-49%) PD-L1 compared with IDH1-/2-mutated iCCA. gLOH, genomic loss of heterozygosity; HRD, homologous recombination deficiency; iCCA, intrahepatic cholangiocarcinoma; TMB, tumor mutational burden; TPS, tumor proportion score.

MSI, TMB, and PDL in IDH-Altered iCCA

To examine potential predictive biomarkers for immunotherapy in IDH-altered iCCA, MSI status, TMB, and PD-L1 status were examined in IDH1mut and IDH2mut iCCA compared with IDHwt (Table 1). Overall, the frequency of MSI-high was low (1%) across all cases of iCCA. IDHwt patients had significantly higher MSI-high status compared with both IDH1- and IDH2-mutated iCCA (P = .009). Similarly, IDHwt patients had significantly higher TMB ≥ 10mut/Mb compared with IDH1- and IDH2-mutated patients (5% of IDHwt patients versus <1% of IDH1- and IDH2-mutated patients, P < .0001; Fig 3B). Of the total population, 847 (34%) patients had PD-L1 status reported. Eighteen percent of patients with IDHwt iCCA were positive for PD-L1 with a range in TPS of 1%-49%. This was significantly higher than IDH1/2mut patients combined (P < .001; Fig 3C). No significant difference was noted between IDH1- and IDH2-mutated iCCA and IDHwt iCCA at higher PD-L1 TPS ≥ 50% (P = .11).

Protein Expression of Immune Microenvironment Markers in IDH-Altered iCCA

To examine markers of the immune microenvironment, 100 surgical tissue samples from 96 patients were subjected to IHC staining for 14 immune microenvironment markers (CD1, CD4, CD8, CD68, PD-1, PD-L1, B7-H3, B7-H4, IDO-1, VISTA, ICOS, OX40, TIM-3, and LAG-2) as previously described.²⁶ In this data set, 29 of 96 (30%) patients were positive for IDH1/2 mutations. Twenty-four (25%) were IDH1mut and five patients (5%) were IDH2mut. No significant difference in immune-microenvironment marker expression was noted between IDH1/2 mutation–positive patients compared with IDHwt (Appendix Fig A1A).

mIF analysis was performed on generated TMA from 84 available patients with iCCA and subjected to two panels of immune markers (Panel 1: CD3, CD8, CD58, PD1, PD-L1, and CK [pan-cytokeratin to identify tumor cells], and Panel 2: CD3, CD8, CD45-RO, FOXP-3, Granzyme B, and CK). No significant difference in colocalization of immune markers or immunophenotypes between IDH1/2 mutation–positive and IDHwt patients was found (Appendix Fig A1B).

RNA Sequencing

RNA sequencing was performed on 73 patients with iCCA, of whom n = 10 had IDH1 mutations, n = 1 had IDH2 mutation, and n = 62 were IDHwt. Given the limited number of IDH2mut cases in this subset, this analysis was focused on IDH1mut compared with non-IDH1mut. Of the 22 immune cells evaluated via CIBERSORT algorithm, resting natural killer (NK) cells were significantly associated with non-IDH1mut compared with IDH1mut (P = .050; Fig 4A). No significant difference was seen in the other cell types, including B cells, dendritic cells, eosinophils, macrophages, mast cells, monocytes, neutrophils, plasma cells, and CD4 and CD8 T cells. Twenty-eight immune checkpoints (CD27, CD274, CD276, CD28, CD70, CD80, CD86, CSF1, CSF1R, CTLA4, FLT3, HHLA2, ICOS, ICOSLG, IDO1, LAG3, NCR3LG1, NT5E, PDCD1, PDCD1LG2, TLR3, TNFRSF18, TNFRSF4, TNFRSF9, TNFSF18, TNFSF4, TNFSF9, and V-set domain containing T-cell activation inhibitor 1 [VTCN1]) were evaluated. Of these, CD70 (P = .037) and programmed cell death 1 (PDCD1; P = .036) were significantly overexpressed in non-IDH1 compared with IDH1mut. VTCN1 expression was significantly higher in IDH1mut than in non-IDH1mut (P = .012; Figs 4B-4D). No significant difference was found between stromal and immune cell infiltration, EMT, vascularization, INF-gamma expression, and T-cell inflammation signature between the two groups.

FIG 4.

RNA sequencing analysis between non-IDH1mut (n = 63) and IDH1mut (n = 10) CCA. (A) Twenty-two immune cell types analyzed using CIBERSORT algorithm showed resting NK cells were significantly associated with non-IDH1mut tumors (P = .05). Analysis of 28 immune checkpoints showed significant overexpression of (B) CD70 (P = .037) and (C) PDCD1 (P = .036) in non-IDH1mut. (D) VTCN1 was significantly overexpressed in IDH1mut (P = .012). CCA, cholangiocarcinoma; CIBERSORT, Cell-type Identification by Estimating Relative Subsets of RNA Transcripts; NK, natural killer; PDCD1, programmed cell death 1; VTCN1, V-set domain containing t-cell activation inhibitor 1.

Survival Analysis of IDH Wildtype and IDH-Mutated Cholangiocarcinoma

Five-hundred one patients with cholangiocarcinoma treated at MD Anderson Cancer Center were retrospectively reviewed (Appendix Table A2). This included 71 IDH1mut and 16 IDH2mut patients. Data from 478 patients were available for survival analysis. The median follow-up time was 33 months (95% CI, 27 to 37). mOS of IDHwt (n = 394) patients was 36 months (95% CI, 29 to 40 months; Fig 5). No significant difference in mOS was detected between the groups with and without IDH mutations (P = .293). No significant difference was noted in mOS between the most prevalent IDH1 isoforms (R132C v R132G/L/K; Appendix Table A3).

FIG 5.

Median OS analysis between IDH-1/2-mutated versus IDHwt patients. ^aP value between IDH-mutated and IDHwt cases. OS, overall survival.

DISCUSSION

IDH1/2 mutations have emerged as targetable alterations in iCCA, and understanding the genomic and immune microenvironment of IDH-altered CCA can lead to improved therapeutic strategies.^14,17 Recently, the mechanisms of acquired resistance to IDH inhibitors have been described, including isoform switching and development of secondary IDH mutations that impair drug binding capacity.^35,36 However, mechanisms of primary resistance to ivosidenib are less well understood and may be related to concurrent mutational profile.⁸

The prevalence of IDH1mut iCCA was 14% and IDH2mut iCCA was 4% (n = 3,067) with a female predilection. This is consistent with previous studies.^3,8,16 In the surgical cases (n = 96), the incidence of IDH1/2 mutations was higher (30%), likely because of smaller sample size. The most frequently mutated IDH1 and IDH2 alleles, respectively, were R132C (69%) and R172K (44%). These are different from the IDH alleles most commonly seen in acute myeloid leukemia (IDH1 R132H and IDH2 R140Q) and in gliomas.^37,38

The tumor suppressor genes ARID1A, BAP1, and PBRM1 were among the most common co-occurring GA in IDH-altered iCCA. PBRM1 mutations are seen in 12% of CCA. PBRM1 loss has been shown to correlate variably in other cancer types in predicting response to immune checkpoint inhibitors (ICIs). In renal cell carcinoma, PBRM1 loss has been associated with ICI response, while in non–small cell lung cancer, it has been linked as a negative predictor for ICI benefit.^39-41 ARID1A, BAP1, and PBRM1 have all been shown to have prognostic significance in CCA, with BAP1 and PBRM1 associated with worse OS and bone metastasis in CCA.²

When compared with IDHwt tumors, the frequency of TP53 mutations, CDKN2A/B, and MTAP loss was significantly lower in IDH1/2mut patients. CDKN2A loss and TP53 mutations have previously been identified as driver mutations in CCA.⁴² Their reduced frequency in IDH-mutated tumors may suggest a less prominent role in oncogenesis of these genes in the presence of IDH alterations. The frequency of targetable alterations (FGFR rearrangements, ERBB2 amplifications, and BRAF mutations) was also significantly lower in IDH1/2mut patients compared with IDHwt, although they are not completely mutually exclusive. Previous transcriptomics studies have subclassified CCA on the basis of mRNA expression.^43,44 Clusters that contained IDHmut CCA were notable for genes regulating mitochondrial function and structure, while such genes were expressed at low levels in FGFR2-fusion containing samples highlighting the genomic heterogeneity of CCA.⁴³

The advent of immunotherapies including ICIs have revolutionized the treatment of many solid malignancies including biliary cancers.^45,46 In the double-blind, placebo-controlled, phase III study TOPAZ-1, the addition of durvalumab (anti–PD-L1 antibody) to gemcitabine/cisplatin was found to significantly improve mOS compared with gemcitabine/cisplatin alone.^47-49 In a subgroup analysis, OS benefit with addition of durvalumab was similar irrespective of presence of IDH1 mutations. Pembrolizumab, when added to gemcitabine and cisplatin, was also found to improve mOS to 12.7 months compared with 10.9 months (HR, 0.83; P = .0034) in KEYNOTE-966.⁵⁰ Although these studies are promising, survival gains remain modest and further research into the immune microenvironment of BTC is needed especially as related to underlying molecular profile.

Recently, Carapeto et al²⁶ characterized the immune landscape of iCCA and identified certain prognostic factors, including central localization high PD-1, LAG3, and low CD3/CD4/ICOS as associated with poor prognosis in patients with CCA. In the current study, we analyzed a subset of these patients with CCA with IDH1/2 mutations. At the protein level, immune microenvironment marker expression and colocalization showed no significant differences in the IDH-mutated cases compared with IDHwt. However, biomarkers of ICI responsiveness from our genomic analysis, including MSI-high status, TMB >10 mut/Mb, and high PD-L1 expression, were significantly less frequent in IDH1mut and IDH2mut iCCA compared with IDHwt, pointing toward an immunologically cold tumor microenvironment of IDH-altered cholangiocarcinoma. Similarly, RNA sequencing data show an increase in resting NK cells, as well as increased expression of the costimulatory molecule CD70 and immune checkpoint PDCD1 in non-IDH1mut iCCA compared with IDH1mut, also implicating IDH1mut tumors to be less immunologically active.^51,52 VTCN1 (B7-H4) expression was higher in the IDH1mut cohort. B7-H4 is a coinhibitory factor that can inhibit CD8+ and CD4+ T-cell growth and cytokine production and has been found to be inversely correlated with PD-L1 in breast cancer.⁵³

Recent research has shown a paucity of immune cells in CCA with reduced cytotoxic T lymphocytes and NK cells and increased immunosuppressive myeloid cells within the tumor-immune microenvironment of CCA.^54-56 In IDH1mut murine models, Wu et al⁵⁷ demonstrated immunoevasion of these tumors by suppression of CD8 T cells and autonomous inactivation of tet methylcytosine dioxygenase 2 DNA demethylase. Inhibition of mutant IDH1 resulted in accumulation of effector CD8+ T cells and produced a phenotype with higher expression of inhibitor checkpoint receptors such as PD-1, CTLA-4, LAG3, and HAVCR2. Combination of ivosidenib with anti–CTLA-4 antibody showed antitumor effect, possibly because of the ability of anti–CTLA-4 therapies to deplete T-regs.⁵⁷

Impaired DNA damage repair has also been seen in IDH-mutated cancer cells.¹⁹ Low levels of NAD+ in IDH1mut cells have shown increased susceptibility to the cytotoxic effects of alkylating agents and PARP inhibitors.^18,58,59 Several ongoing studies are looking into PARP inhibitors, alone or in combination with immunotherapy, in IDH-mutated solid malignancies including cholangiocarcinoma (ClinicalTrials.gov identifier: NCT03991832, NCT03212274). In the current analysis, g(LOH) was used as a surrogate biomarker for HRD. The LoH threshold evaluated in this study would be hypothesis-generating for potential use of IDH inhibitors to target DNA damage response; however, no significant difference was found in gLOH in IDHmut compared with IDHwt iCCA population.

To complement the genomic profiling, a retrospective analysis was performed of 501 patients with cholangiocarcinoma and showed no significant difference in mOS between IDH1, IDH2, and IDHwt patients, consistent with previous reports.¹⁶ Both early- and late-stage patients were included in the survival analysis, which likely contributed to the longer mOS (38 months) in this cohort. Additionally, patients were provided multidisciplinary care with 29% of patients enrolled in clinical trials, which may have also improved outcomes.

In conclusion, IDH-mutated CCA had a lower frequency of other known driver mutations compared with IDHwt reflecting its driver oncogenic status. IDH-mutated CCA tends to be more immunologically cold and without significant differences in g(LOH) compared with IDHwt CCA. These data are expected to be instructive when planning precision immuno-oncology therapies in CCA.

APPENDIX

FIG A1.

IHC and multiplex immunofluorescence (mIF) analysis. (A) Representative pictomicrographs of 4 markers (PD-L1, CD3, CD8, and B7-H4) taken at 200-µm magnification shown of IHC analysis. A total of 14 immune markers examined by IHC in 96 patients with iCCA showed no significant difference between IDH1/2mut and IDHwt. (B) Representative pictomicrographs of IDHwt, IDH1, and IDH2 mutation-positive patients from constructed tissue microarrays subjected to mIF analysis. H&E stain, panel 1, and panel 2 mIF images at 20× magnification showed no significant difference between IDH1/2mut and IDHwt in immune panel expression or colocalization. iCCA, intrahepatic cholangiocarcinoma; H&E, hematoxylin and eosin; IHC, immunohistochemistry.

TABLE A1.

Co-occurring Genomic Alterations With IDH1/2 Mutation Status

Co-Occurring Mutation	IDH1+ iCCA	IDH1– iCCA	IDH2+ iCCA	IDH2– iCCA
Total cases, No.	426	2,641	125	2,942
TP53, %	12	38	9	35
CDKN2A, %	20	32	12	31
CDKN2B, %	16	23	9	22
KRAS (non-G12C), %	6	21	2	20
MTAP, %	8	16	6	15
BAP1, %	17	13	23	14
TERT, %	0.2	8	2	7
SMAD4, %	1	8	1	8
MYC, %	1	5	0	5
FGFR2 fusions, %	1	10	1	9
ERBB2, %	2	5	2	5
BRAF, %	4	5	2	5
PIK3CA, %	8	6	3	7
PTEN, %	1	3	2	3
KRAS G12C, %	0.20	1	0	1
ARID1A, %	25	18	19	19
PBRM1, %	19	10	14	11
STK11, %	0.2	3	1	3
MDM2, %	1	5	2	4
CD274 amplification, %	0	0.30	1	0.20

Abbreviation: iCCA, intrahepatic cholangiocarcinoma.

TABLE A2.

Clinical Characteristics of Patients With CCA Treated at MD Anderson Cancer Center Between 2008 and 2022 on the Basis of IDH1/2 Mutational Status

Characteristic	IDH1 GA (n = 71)	IDH2 GA (n = 16)	IDH WT (n = 415)
Age, years, median (range)	64 (30-83)	60 (31-79)	61 (21-87)
Sex, No. (%)
Male	27 (38)	6 (37.5)	196 (47.3)
Female	44 (62)	10 (62.5)	218 (52.6)
Stage, No. (%)
I or II	14 (19.7)	4 (25)	69 (16.7)
III or IV	41 (57.7)	8 (50)	345 (83.3)
Unknown	16 (25.5)	4 (25)
Subtype, No. (%)
Intrahepatic	70 (98.5)	16 (100)	414 (100)
Extrahepatic	1 (1.5)	0	1 (0.2)
Median number of systemic therapies, No. (%)	2 (1-7)	2 (0-7)	2 (1-12)
Treated on clinical trial, No. (%)	23 (32.4)	6 (37.5)	113 (27.2)
Treated with IDH1 inhibitor, No. (%)	23 (32.4)	0	0
Received ChemoXRT, No. (%)	12 (17)	2 (12.5)	59 (14.3)
Received surgery, No. (%)	19 (26.7)	1 (6.3)	53 (12.8)
Received liver-directed therapy, No. (%)	27 (38)	5 (31.3)	128 (31)

Abbreviations: CCA, cholangiocarcinoma; GA, genomic alteration.

TABLE A3.

Median OS Analysis Between IDH1 Isoforms

IDH1 Isoform	No.	Median OS, Months (95% CI)	Log-Rank Test P Value
R132C	47	38 (24 to NA)	.6
Othera	33	46 (22 to NA)
R132C	47	38 (24 to NA)	.1
R132G	7	22 (17 to NA)
R132L	8	10 (7 to NA)
R172K	6	NA (12 to NA)

Abbreviations: NA, not applicable; OS, overall survival.

^a Other indicates R132G/L/K combined.

AUTHOR CONTRIBUTIONS

Conception and design: Shalini Makawita, Lawrence N. Kwong, Anil Korkut, Jeffrey S. Ross, Milind Javle

Collection and assembly of data: Shalini Makawita, Lawrence N. Kwong, Zeyad Abouelfetouh, Anaemy Danner De Armas, Karthikeyan Murugesan, Natalie Danziger, Dean Pavlick, Jeffrey S. Ross, Milind Javle

Data analysis and interpretation: Shalini Makawita, Sunyoung Lee, Elisabeth Kong, Lawrence N. Kwong, Lianchun Xiao, Karthikeyan Murugesan, Natalie Danziger, Dean Pavlick, Anil Korkut, Jeffrey S. Ross, Milind Javle

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/po/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).