Measurable Residual Disease Monitoring by Locked Nucleic Acid Quantitative Real-Time PCR Assay for IDH1/2 Mutation in Adult AML

Simple Summary Measurable residual disease (MRD) monitoring is crucial in managing AML to predict the risk of relapse. A better understanding of which MRD technique and molecular target will have an effective clinical impact on AML is still required. Locked nucleic acid quantitative Real-Time PCR assay (LNA-qPCR) is sensitive and specific for quantifying oncogenetic single-nucleotide variation. We assessed the role of LNA-qPCR in the monitoring of IDH1/2 mutations MRD in eighty-eight AML patients from multiple centers. We found that IDH1/2 LNA-qPCR MRD correlates well with NPM1 qPCR MRD, predicts relapse-free survival and cumulative incidence of relapse, and is a potential MRD technique for IDH1/2-mutated AML patients with reduced IDH1/2 mutant levels after complete remission. Abstract Locked nucleic acid quantitative Real-Time PCR (LNA-qPCR) for IDH1/2 mutations in AML measurable residual disease (MRD) detection is rarely reported. LNA-qPCR was applied to quantify IDH1/2 mutants MRD kinetics in bone marrow from 88 IDH1/2-mutated AML patients, and correlated with NPM1-MRD, clinical characteristics, and outcomes. The median normalized copy number (NCN) of IDH1/2 mutants decreased significantly from 53,228 (range 87–980,686)/ALB × 106 at diagnosis to 773 (range 1.5–103,600)/ALB × 106 at first complete remission (CR). IDH1/2 LNA-qPCR MRD was concordant with remission status or NPM1-MRD in 79.5% (70/88) of patients. Younger patients and patients with FLT3 mutations had higher concordance. The Spearman correlation coefficient (rs) and concordance rate between the log reduction of IDH1/2 LNA-qPCR and NPM1-MRD were 0.68 and 81% (K = 0.63, 95% CI 0.50–0.74), respectively. IDH1/2-MRD > 2 log reduction at first CR predicted significantly better relapse-free survival (3-year RFS rates 52.9% vs. 31.9%, p = 0.007) and cumulative incidence of relapse (3-year CIR rates 44.5% vs. 64.5%, p = 0.012) compared to IDH1/2-MRD ≤ 2 log reduction. IDH1/2-MRD > 2 log reduction during consolidation is also associated with a significantly lower CIR rate than IDH1/2-MRD ≤ 2 log reduction (3-year CIR rates 42.3% vs. 68.8%, p = 0.019). LNA-qPCR for IDH1/2 mutation is a potential MRD technique to predict relapse in IDH1/2-mutated AML patients, especially for those with IDH1/2 MRD > 2 log reduction at first CR or a concurrent FLT3 mutation.


Introduction
Acute myeloid leukemia (AML) is a hematological malignancy characterized by clonal expansion of myeloid progenitors in the bone marrow (BM), peripheral blood, and other tissues. AML is a heterogeneous disease in terms of molecular genetics and phenotypic characteristics [1,2]. Molecular screening plays a significant role in prognostic categorization and decision of treatment strategies for AML [3].
Minimal or measurable residual disease (MRD) has prognostic value in AML patients during or after treatment [14][15][16][17]. The use of IDH1/2 mutations as molecular markers for MRD monitoring is still under investigation. A few studies have reported the stability and suitability of IDH1/2 mutations to monitor MRD by digital PCR (ddPCR) [18,19], nextgeneration sequencing (NGS) [20][21][22], and locked nucleic acid (LNA) quantitative Real-Time PCR (LNA-qPCR) [23,24]. Persistent IDH1 or IDH2 mutations have been observed in AML patients at the time of morphological remission [22,25]. Two studies with large cohorts of patients have reported that persistent mutated IDH1/2 MRD detected by ddPCR or NGS in AML at the time of complete remission (CR) could predict relapse [19,20]. The impact of IDH1/2 MRD by LNA-qPCR on the outcome of AML patients has not been investigated in large cohorts previously.
NPM1 mutation is a stable MRD marker and NPM1 PCR assays are recommended in AML patients with NPM1 mutation [14,17,26]. In the present study, we aimed to evaluate the use of LNA-qPCR assay to monitor MRD after treatment in AML patients harboring IDH1-R132, IDH2-R140, or IDH2-R172 mutations, to correlate the MRD results of IDH1/2 mutations with those of NPM1 mutations, and to determine their impact on the outcome.

Patients and Samples
Newly diagnosed adult de novo AML patients who received induction therapy, including standard chemotherapy, azacitidine, or low-dose cytarabine, were enrolled. This study was conducted by the AML consortium of Taiwan, which aimed to detect genetic mutations and monitor MRD for AML patients in Taiwan. BM samples at the initial diagnosis were analyzed for cytogenetics and a panel of gene mutations. G-banding method was used for karyotypic analysis. AML classification and risk categorization followed the criteria of 2022 WHO classification and 2022 ELN recommendations [3,27]. The study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (99-0112B, 102-3031B, 201700154B0, and 201701453A3) and the participating hospitals (CMUH-HO-AML-10701). Informed consent was obtained from enrolled patients.
BM samples were collected at diagnosis and at different time points after treatment, including the first CR documented after induction therapy, during consolidation (after one to two cycles of consolidation therapy), end of treatment, and relapse. Mononuclear cells were enriched by using Ficoll density gradient centrifugation and cryopreserved. Genomic DNA (gDNA) or RNA extraction and cDNA preparation were performed as previously described [28].

Monitoring of IDH1/2 MRD by LNA-qPCR Assay
Quantification of IDH1/2 mutation copy number was assessed by quantitative Real-Time PCR (qPCR) for gDNA using the TaqMan technology with LNA probes specific for the mutant allele to increase the specificity for single-nucleotide variation in IDH1-R132, IDH2-R140, and IDH2-R172. The primer sequences and DNA probes used to amplify IDH1-R132, IDH2-R140, and IDH2-R172 mutations are shown in Table 1. Probes, forward primers, and reverse primers for IDH2-R140Q, IDH2-R140W, IDH1-R132H, IDH1-R132S, IDH1-R132G, and ALB were the same as those previously reported [24,29]. The reverse primer of IDH2-R172K was modified by removing the R-LNA. Reverse primers of IDH1-R132C and IDH1-R132L were designed in our lab. A locked nucleic acid was fixed in the last 3 nucleotide of the reverse primers. Reaction mixtures of 25 µL contained the 100 ng gDNA, 2 × TaqMan Master Mix buffer (ThermoFisher), 0.08% BSA, 400 nM of each primer, and 200 nM probe. The PCR cycling protocol consisted of 2 min at 50 • C, 10 min of initial denaturation at 92 • C, and was followed by 50 cycles of 15 s at 95 • C, 1 min at 58 • C. Albumin (ALB) was used as a reference gene. IDH1/2 standard curves were constructed by serial dilution of IDH1-R132, IDH2-R140, and IDH2-R172 plasmids with wild-type IDH1/2 gDNA from normal peripheral blood (Figure 1a-d). Specificity was determined by negative control with wild-type IDH1/2 gDNA. The sensitivity of the assay was five copies for IDH1-R132C/S/G/L, ten copies for IDH1-R132H, five copies for IDH2-R140Q/W, and ten copies for IDH2-R172K with a quantitative range from 5-100 to 10 6 copies. The normalized copy number (NCN) of the IDH1/2 mutant of the sample was determined by the IDH1/2-mutant copies/ALB × 10 6 copies. The result of IDH1/2 MRD was expressed as log reduction by calculation of NCN follow-up divided by NCN diagnosis for the individual patients. The mean NCN (IDH1/2 copies/ALB × 10 6 copies) of 19 (Table  S1, Figure S1). The MRD negativity is defined as a NCN level that is less than the mean NCN of normal BM samples for each IDH1/2 mutation type.  GTCGCCATGGGCGTGCT  TCCCACGCCTAGTCCCTGGCTG  FAM-AGCCCATCACCAT-MGB   ALB  CTCTCCTTCTCAGAAAGTGTGCATAT  TGAAACATACGTTCCCAAAGAGTTT  FAM-TGCTGAAACATTCA  CCTTCCATGCAGA -TAMRA R-LNA-0, a locked nucleic acid was fixed in the last 3 nucleotide of the reverse primers.

Detection of NPM1 RT-qPCR MRD
NPM1 MRD was detected by quantitative Real-Time reverse transcriptase PCR (RT-qPCR) with the TaqMan assay using ABL1 as the internal control gene. Primers and probes used were according to the method described by Gorello et al. [30]. The MRD results were expressed as a log reduction of the NCN at follow-up compared to the NCN at diagnosis.

Statistical Analysis
Wilcoxon's rank-sum test, Fisher's exact test, and the χ 2 analysis were used whenever appropriate to make comparisons between groups. The correlation between IDH1/2 LNA-qPCR, marrow blasts, and NPM1 RT-qPCR was assessed by Spearman's rank correlation test. Cohen's kappa (K) was used to evaluate the concordance between IDH1/2 LNA-qPCR and NPM1 RT-qPCR. The IDH1/2 MRD at first CR after induction therapy and during consolidation was subjected for outcome analysis. The overall survival (OS), relapse-free survival (RFS), and cumulative incidence of relapse (CIR) are based on the 2022 ELN AML recommendation [3]. Estimates of survival were calculated according to the Kaplan-Meier method. The log-rank test analyzed the differences in survival for significance. Differences in the CIR were assessed using the Gray test. In all analyses, the p-values were two-sided and considered statistically significant when values < 0.05. Statistical analysis was carried out by R version 4.1.3, R Foundation for Statistical Computing, Vienna, Austria.

Correlation of IDH1/2 LNA-qPCR MRD with Clinical and Genetic Characteristics of AML Patients
The difference in MRD results between NPM1 RT-qPCR MRD and IDH1/2 LNA-qPCR methods was analyzed for patients with good correlations and IDH1/2 LNA-qPCR MRD > 2 log reduction. The median log reduction of MRD was 5 log (range 2.7-5.0) by NPM1 RT-qPCR and 2.9 (range 2.0-4.6) by IDH1/2 LNA-qPCR. The level of MRD log reduction was significantly higher by NPM1 RT-qPCR than by IDH1/2 LNA-qPCR
Patients with a good correlation between IDH1/2 LNA-qPCR MRD and AML remission status were significantly younger than patients without correlation (median age 53 vs. 61 years, p = 0.025). Patients with FLT3 mutations (either FLT3-ITD or FLT3-TKD) had significantly higher concordance rates between IDH1/2 LNA-qPCR MRD and AML remission status than patients without FLT3 mutations (96% vs. 71%, p = 0.012). The diagnostic IDH1/2 NCN, white blood cell counts, hemoglobin, platelet counts, percentages of circulating or marrow blasts, and the mutation status of NPM1 had no effect on the correlation between IDH1/2 LNA-qPCR MRD and AML remission status (Table 3).
To explore the clonal evolution of IDH1/2 mutated AML patients in the different stages, we detected co-mutations with allele frequencies in one patient with discordant IDH1-R132H MRD and another patient with concordant IDH2-R140Q MRD at the diagnosis, CR, and relapse samples. The patient with discordant IDH1-R132H MRD had stable clone sizes of DNMT3A mutation and IDH2-R140Q at diagnosis, CR, and relapse, but lost STAG2 mutation and acquired CEBPA mutation at relapse (Figure 5a). Another patient with concordant IDH2-R140Q had reduced allele frequency of IDH2-R140Q and other comutations at CR, but IDH2-R140Q and co-mutations at diagnosis re-expanded at relapse (Figure 5b). Patients with a good correlation between IDH1/2 LNA-qPCR MRD and AML remission status were significantly younger than patients without correlation (median age 53 vs. 61 years, p = 0.025). Patients with FLT3 mutations (either FLT3-ITD or FLT3-TKD) had significantly higher concordance rates between IDH1/2 LNA-qPCR MRD and AML remission status than patients without FLT3 mutations (96% vs. 71%, p = 0.012). The diagnostic IDH1/2 NCN, white blood cell counts, hemoglobin, platelet counts, percentages of circulating or marrow blasts, and the mutation status of NPM1 had no effect on the correlation between IDH1/2 LNA-qPCR MRD and AML remission status (Table 3).
To explore the clonal evolution of IDH1/2 mutated AML patients in the different stages, we detected co-mutations with allele frequencies in one patient with discordant IDH1-R132H MRD and another patient with concordant IDH2-R140Q MRD at the diagnosis, CR, and relapse samples. The patient with discordant IDH1-R132H MRD had stable clone sizes of DNMT3A mutation and IDH2-R140Q at diagnosis, CR, and relapse, but lost STAG2 mutation and acquired CEBPA mutation at relapse (Figure 5a). Another patient with concordant IDH2-R140Q had reduced allele frequency of IDH2-R140Q and other co-mutations at CR, but IDH2-R140Q and co-mutations at diagnosis re-expanded at relapse (Figure 5b).

Discussion
Our cohort of 88 adult patients with IDH1/2-mutated AML showed that the concordance rate of IDH1/2 LNA-qPCR MRD with remission status was 79.5%. There was a positive correlation (r s = 0.68) and a substantial agreement (81%, K = 0.74) between IDH1/2 LNA-qPCR and NPM1 RT-qPCR MRD. The median difference in MRD results was 2 logs between IDH1/2 LNA-qPCR and NPM1 RT-qPCR. Patients with FLT3 mutation or younger age have a higher concordance rate between IDH1/2 LNA-qPCR MRD, remission status, or NPM1 RT-qPCR MRD. At the end of induction and during consolidation, IDH1/2 LNA-qPCR MRD had a significant impact on CIR or RFS.
Because of the single nucleotide substitution of IDH1/2 mutations, the sensitivity and specificity of IDH1/2-directed MRD are limited if there is a small proportion of IDH1/2 mutations in the wild-type background. IDH1/2-targeted MRD using ddPCR assays reached a sensitivity of 0.1% due to the problem of a relatively high background observed in negative controls, which consisted of false-positive droplets [19]. LNA primers could block the wild-type allele, avoid the false-positive signals resulting from polymerase errors during PCR amplification steps, and allow for higher sensitivity and specificity of IDH1/2 LNA-qPCR [24,31]. Previous studies have applied diverse methods to detect IDH1/2-targeted residual disease by using NGS, ddPCR, or LNA-qPCR. The sensitivity of IDH1/2 MRD was reported to be 0.1%-0.2% by ddPCR [18,19,32], < 0.001%-1% by NGS [20][21][22], and 10 −4 by LNA-qPCR [24]. LNA-qPCR assay for IDH1/2 mutations theoretically has a sensitivity of 1 × 10 −4 , allowing for a deeper evaluation of molecular response. However, we observed a low copy number background of IDH1/2 mutations by LNA-qPCR in normal BM samples, especially for IDH1-R132C (c.394C > T) and IDH1-R132H (c.395G > A). There was also a higher mean variant allele fraction of IDH1-R132C and IDH1-R132H than other IDH1/2 mutant types detected by ddPCR assay on gDNA extracted from IDH1/2 wild-type pooled blood lymphocytes in the study by Ferret et al. [19]. The low copy number background of IDH1/2 LNA-qPCR in normal BM samples seems to be related to the mutant subtype rather than age ( Figure S1). Clonal hematopoiesis is unlikely to be the cause of the low copy number background of IDH1/2 mutants. A possible explanation for this background noise is PCR-induced substitution error caused by cytosine deamination and increased C > T and G > A transition errors, which exhibit strong sequence context dependency [33,34]. We observed a persistently high level of IDH1/2 mutation at CR in 20% of IDH1/2mutated patients. The discrepancy of persistent IDH1/2 LNA-qPCR MRD in CR might be attributed to clonal hematopoiesis in pre-leukemic hematopoietic stem cells, which provide a potential reservoir for leukemic progression [35,36]. This hypothesis is supported by the observation of persistent IDH1-R132H and DNMT3A mutation, loss of STAG2 mutation at CR, and acquisition of CEBPA mutation in one patient in this study. Previous studies reported that persistent IDH1/2 mutants in AML patients at CR occurred in 39-62% by NGS and 7% by ddPCR [19][20][21]. We found that NPM1 RT-qPCR MRD showed a median difference of two log reductions compared with IDH1/2 LNA-qPCR MRD in AML. Low copy number false-positive IDH1/2 LNA-qPCR background due to PCR-induced substitution errors of IDH1/2 mutant and detection of gDNA by IDH1/2 LNA-qPCR rather than cDNA by NPM1 RT-qPCR, where a single cell may contain multiple copies of RNAs, and low NCN of IDH1/2 mutants at initial diagnosis might all contribute to the limited sensitivity of IDH1/2 LNA-qPCR. Due to this, RT-qPCR MRD targeting NPM1 mutant rather than IDH1/2 mutant is recommended for AML patients with both NPM1 and IDH1/2 mutations.
The use of FLT3 mutations as a genetic marker for MRD is controversial. This is due to unstable FLT3 mutations that are acquired or lost during the disease course of AML [37,38]. It lacks standardization of the detection methods for FLT3 MRD regarding mutation types covered and limited sensitivity [39][40][41]. The ELN MRD working party recommends that the use of FLT3 mutations is appropriate in combination with other additional MRD markers [26]. NGS-MRD for multiple genetic markers is not widely available. Only one of 25 (4.0%) AML patients with concurrent IDH1/2 and FLT3 mutations had a discrepancy between IDH1/2 mutant MRD and remission-relapse status. Based on this observation, IDH1/2 mutation might be a helpful surrogate or additional genetic marker for MRD monitoring in patients with concurrent FLT3 mutation and wild-type NPM1.
The loss of IDH1/2 mutation at relapse of AML has been rarely reported. Brambat et al. reported the loss of IDH2 mutation at relapse in one of 14 post-transplantation AML patients [18]. All IDH1/2-mutated AML patients who had achieved CR retained IDH1/2 mutant at relapse. The small number of cases detected for IDH1/2-MRD at relapse in this cohort might preclude the finding of IDH1/2 disappearing at relapse. Whether IDH1/2 mutation could be a stable genetic MRD marker will need to be validated in a larger cohort. The stability of NPM1 mutation, a consensus recommended MRD marker for AML, was reported to be 99%-100% in large cohorts [14,42], but the loss of NPM1 mutation at relapse has been observed in 7% to 25% of AML patients in smaller series [17,[43][44][45][46][47]. Because of the marked genetic heterogeneity of AML, no single genetic MRD marker can be applied to all patients. As the role of MRD in the management of AML patients is becoming more and more important, IDH1/2 mutation might be a surrogate MRD marker for a subset of AML patients without recurrent fusion genes or NPM1 mutations.
A large cohort study of the French ALFA group reported that IDH1/2 mutant allele burden < 0.2% by ddPCR after induction was a significant predictor of prolonged eventfree survival [19]. The MD Anderson Cancer Center study showed that persistent IDH1/2 mutation in remission detected by NGS at a sensitivity of 1% predicted relapse in multivariate analyses [20]. The results of this study further support the impact of IDH1/2 MRD on relapse. In a recent study reported by Bill et al., IDH1-R132 and IDH2-R172 MRD positivity in remission at HSCT was associated with an increased risk of relapse, while IDH2-R140 mutations did not [48]. We observed that IDH1/2 LNA-qPCR MRD elevations preceded frank hematological relapse and confirmed that IDH1/2 LNA-qPCR MRD after induction therapy and during consolidation could predict relapse in AML patients but did not significantly predict OS. However, we found that IDH2-R140 MRD positivity postinduction and during consolidation could predict an increased risk of relapse, whereas IDH1-R132 or IDH2-R172 did not.
Studies showed that early pre-emptive interventions might improve the outcomes of AML patients in CR with persistent positive fusion genes or NPM1 mutation [49,50]. However, the feasibility, outcome prediction, and guidance of pre-emptive therapy based on IDH1/2-MRD in AML patients was rarely investigated and thus cannot be recommended, mainly due to the concern of IDH1/2 mutants as pre-leukemic clones as well as the possibility of loss or gain of IDH1/2 mutants at relapse [26]. For patients with IDH1/2 MRD > 2 log reduction at the end of induction, the IDH1/2 MRD correlated with NPM1 MRD and morphological remission status, and the loss of IDH1/2 mutation at relapse rarely occurred in the current and previous studies [18,19]. Patients with a deep IDH1/2 molecular response after achieving CR will benefit from serial IDH1/2 MRD monitoring to identify early relapse. On the other hand, patients with persistently high IDH1/2 MRD at first CR should not be followed by IDH1/2-targeted MRD alone. It is still unclear whether early pre-emptive therapy will benefit AML patients in CR who have increasing IDH1/2 mutant MRD during follow-up, which requires further evaluation in prospective clinical trials.

Conclusions
Our study showed that the correlation and stability of IDH1/2 MRD by LNA-qPCR was good in patients who had concordantly reduced IDH1/2 MRD in morphologic CR. IDH1/2 LNA-qPCR MRD > 2 log reduction at first CR and during consolidation is associated with improved CIR or RFS. IDH1/2 LNA-qPCR MRD has the potential to become a clinical tool to monitor deep treatment response and predict relapse for IDH1/2 mutated AML patients, especially for those who are younger, with reduced IDH1/2 MRD at first CR, or concurrent FLT3 mutation.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cancers14246205/s1, Table S1. IDH1/2 LNA-qPCR of 19 normal bone marrow samples from healthy donors. Figure S1. IDH1/2 LNA-qPCR of 19 normal bone marrow samples from healthy donors according to age. Figure S2. MRD kinetics of individual patients with good correlation between IDH1 LNA-qPCR MRD, NPM1 RT-qPCR, or remission-relapse status. Figure S3. MRD kinetics of individual patients with discordant IDH1 LNA-qPCR MRD, NPM1 RT-qPCR MRD, or remission status. Figure S4. MRD kinetics of individual patients with good correlation between IDH1 LNA-qPCR MRD and hematological remission-relapse status. Figure S5. MRD kinetics of individual patients with persistent high IDH1 LNA-qPCR and relapse refractory status. Figure S6. MRD kinetics of individual patients with good correlation between IDH2 LNA-qPCR MRD, NPM1 RT-qPCR, or remission-relapse status. Figure S7. MRD kinetics of individual patients with discordant IDH2 LNA-qPCR MRD, NPM1 RT-qPCR MRD, and hematological remission status. Figure S8. MRD kinetics of individual patients with good correlation between IDH2 LNA-qPCR MRD and hematological remission-relapse status. Figure S9. MRD kinetics of individual patients with persistent high IDH2 LNA-qPCR and relapse refractory status. Figure S10. Summary of MRD kinetics according to the concordance between IDH1/2 LNA-qPCR MRD and the disease status.  Informed Consent Statement: Informed consent was obtained from enrolled patients.

Data Availability Statement:
The datasets used and/or analyzed in the current study are available from the corresponding author on reasonable request.