3.1 Study Selection

In this systematic review, 21 studies were included that measured DNA-TG levels in WBC for either patients with ALL (n = 16) or IBD (n = 5). Furthermore, the studies, which investigated different methods for measuring DNA-TG, are discussed here below in greater detail. A flowchart of the included studies can be found in Fig. 4.

Fig. 4

PRISMA flowchart of included studies in this systematic review [25]

3.2 Analytical Methods for Measuring DNA-TG and 6-TGNs

In the late 1970s, initial attempts were made to measure DNA-TG. Tidd and Dedhar (1978) [27] used a high-performance liquid chromatography (HPLC) technique based on DNA hydrolysis and quantitation of a fluorescent oxidation product of 6-TG by anion-exchange HPLC in leukocytes. However, this method was not suitable for therapeutic drug monitoring (TDM) due to its low sensitivity and large quantities of pure DNA required for analysis. From the early 1990s to the late 2000s, various studies were published that involved quantification of TG in the DNA of leukocytes [28,29,30,31,32]. In all these studies, DNA was isolated from peripheral blood and digested using nucleases and phosphatases, resulting in free deoxynucleosides which could be detected using HPLC with UV absorption or fluorescence detection. In 2012, Jacobsen et al. [24] published a novel method to quantify DNA-TG using liquid chromatography-tandem mass spectrometry (LC–MS/MS). Coulthard et al. [33] also described a method of measuring DNA-TG levels in thiopurine-treated patients with IBD. The difference in these latter studies was that the method of Jacobsen, but not Coulthard, required derivatization before analysis, necessitating standard dTG-incorporated DNA to control for variation in the derivatization reaction. In 2021, Choi et al. [23] published a modified assay to measure DNA-TG levels in leukocytes by LC–MS/MS, which they tested in patients with ALL. The advantage of their method is that it required—and was validated with—only small volumes of DNA (1 µg), making it potentially suitable for routine use of DNA-TG measurements in clinical practice.

Currently, measurement of RBC 6-TGNs is the most widely used thiopurine therapeutic drug monitoring. The assays of Lennard et al. [34] and Dervieux [35] are the most commonly utilized in practice, with the Dervieux [35] assay being widely adopted due to its slightly lower coefficient of variation and greater simplicity. Results and therapeutic ranges from the two assays cannot be directly compared, because RBC 6-TGN levels with the Dervieux [35] assay are 2.6-fold higher than those obtained by the Lennard assay [36]. The analytical methods used in the studies covered by this review are provided in Table 2 [23, 24, 33]. A meta-analysis of WBC DNA-TG/RBC 6-TGN correlation coefficients where provided, using the Fisher’s z-transformed correlation coefficient for meta-analysis was performed (and is shown in Fig. 2). In a fixed effects model, the overall Fisher’s z-transformed correlation coefficient was 0.59 [95% confidence interval (CI) 0.54–0.64], consistent with a moderate to strong correlation between WBC DNA-TG and RBC 6-TGN levels.

Table 2 Overview of the various analytical methods that have been used for measuring DNA-TG and TGNs3.3 DNA-TG in Relationship to Efficacy and Safety3.3.1 Relapse-Free Survival in Acute Lymphoblastic Leukemia

Nielsen et al. [37] investigated the association of DNA-TG, RBC 6-TGNs, and other metabolites with relapse-free survival within the NOPHO ALL2008 trial for patients with acute lymphoblastic leukemia (ALL). The treatment protocol consisted of maintenance therapy with MP 75 mg/m2 once per day and methotrexate 20 mg/m2 once per week. It was found that relapse-free survival was significantly associated with DNA-TG concentrations [adjusted HR (HRa) 0.81 per 100 fmol/µg DNA increase; 95% CI 0.67–0.98]. Hence, relapse-free survival in patients with ALL increased with increasing DNA-TG concentrations. In comparison, relapse-free survival was not associated with RBC 6-TGN (adjusted HR 0.96 per 100 nmol/mmol hemoglobin increase, 95% CI 0.80–1.27). In an attempt to further optimize the abovementioned maintenance therapy, Larsen et al. [38] added incremental low doses of TG to the treatment protocol (TEAM protocol, starting TG dose 2.5 mg/m2/day to maximum dose 12.5 mg/m2/day) [39]. It was found that addition of TG to the treatment protocol led to higher levels of DNA-TG compared with historical patients who received only standard therapy of methotrexate and MP. Unsurprisingly, higher 6-TGN levels were found in patients in whom TG was added to the treatment protocol, but addition of TG did not lead to increased myelotoxicity or hepatotoxicity. The superiority of adding TG to conventional antileukemic 6MP/methotrexate maintenance therapy in pediatric ALL is currently being tested in an international randomized clinical trial [40]. The association between DNA-TG concentrations and the relapse rate in acute lymphoblastic leukemia (ALL) was also studied by Toksvang et al. [41] in an individual participant data meta-analysis (IPD-MA). In their IPD-MA, they included six trials that involved 1901 pediatric and young adult patients with ALL treated with MP (start dose, 75 mg/m2/day). The median DNA-TG level was 415 fmol/μg [interquartile range (IQR) 282–587, n = 1609] and did not differ between males and females, age groups, patients with white blood cell count < 50 versus ≥ 50 × 109/L at diagnosis. The median follow-up time was 6.0 years (IQR 4.1–8.4, n = 1910). In the pooled data-analysis, the relapse-specific HRa was 0.94 per 100 fmol/μg increase in DNA-TG (95% CI 0.88–1.00, n = 1910). In patients with end-of-induction minimal residual disease (EOI MRD)-positive patients (n = 839), the HRa’s were 0.87 (95% CI 0.78–0.97) and 0.90 (95% CI 0.82–0.99) per 100 fmol/μg increase in DNA-TG for relapse and any event (relapse, second cancer, or death). DNA-TG levels were not associated with relapse or any event in EOI MRD-negative patients. Hence, this IPD-MA showed that DNA-TG is inversely associated with the relapse rate in patients with ALL, especially in EOI MRD-positive patients with ALL. As the authors mentioned, the latter is particularly encouraging as EOI MRD is a very important prognostic marker for event-free survival in pediatric and young adult patients with ALL. This also supports the investigation and utility of clinical monitoring of DNA-TG concentrations in patients with ALL.

3.3.2 Leukopenia in Patients with ALL

Leukopenia has been described in various studies using DNA-TG levels in patients with ALL. Ju et al. [42] reported that 8 out of 63 patients who underwent MP maintenance therapy for ALL developed leukopenia (WBC < 1500/μL) at the time of DNA-TG sampling. Four patients had NUDT15 variants (two *1/*2, one *1/*5, and one *1/*6 genotype), while no patients had TPMT variants. In their study, the DNA-TG levels ranged from 27.8 to 504.8 fmol TG/μg DNA. Furthermore, no correlation was found between DNA-TG (P = 0.093) or 6-TGN (P = 0.425) levels with WBC counts. Fan et al. [43] reported that patients with NUDT15 variants (c.36_37insGGAGTC, c.352G > A, c.415C > T) experienced significantly higher episodes of leukopenia during therapy. Patients with intermediate or low NUDT15 activity showed a higher risk of developing leukopenia compared with wild-type activity (odds ratio of 6.00, P = 0.009). However, no statistically significant correlation was observed between DNA-TG levels and lowest WBC and lowest absolute neutrophil count. Although, a significantly higher DNA-TG/MP dose ratio was found in patients who experienced one or more leukopenia episodes (P = 0.025, n = 16). In our meta-analysis, no significant difference (mean difference is 57.7 (95% CI − 22.93 to 138.35, P = 0.16) was found between DNA-TG levels of patients with ALL who developed leukopenia compared with those who did not develop leukopenia. The explanation why no association between DNA-TG and leukopenia was found in patients with ALL might be due to various reasons such as use of concomitant methotrexate [43], no correction for thiopurine dosage, or NUDT15/TPMT/other variants, and lack of 6-MMP measurements.

3.3.3 DNA-TG Measurements Predict Late Leukopenia in Inflammatory Bowel Disease

Two studies were performed that compared DNA-TG and 6-TGN measurements in predicting leukopenia in patients with IBD [44]. The first study was performed by Zhu et al. [44] who used DNA-TG and 6-TGN measurements at time of late leukopenia or after 8 weeks after stable dose was obtained and investigated whether these measurements were able to predict late leukopenia (developed after 2 months) during a prospective follow-up of patients. It was found that patients with IBD who developed late leukopenia (> 2 months) had significantly higher DNA-TG levels compared with patients who did not develop late leukopenia (423.3, IQR 361.1–577.4 versus 270.0; IQR 188.1–392.4 fmol/µg DNA). However, no significant differences in RBC 6-TGN concentrations were found between patients with IBD who developed late leukopenia compared with patients with IBD who did not develop late leukopenia.

With respect to DNA-TG, these investigators used a cutoff level of 340.1 fmol/µg DNA for the prediction of late leukopenia (sensitivity 83.7% and specificity 63.4%; positive predictive value 27.1% and negative predictive value 96.0%). Interestingly, they also compared the abovementioned DNA-TG cutoff level with a 6-TGN cutoff level (≤ 355 pmol/8 × 108 RBC) from literature based on patients from Asian descent [45]. The results showed that of IBD patients that developed leukopenia who were deemed to have “normal” 6-TGN-levels, 56.5% of them had “high” DNA-TG levels. DNA-TG was also associated with late leukopenia in NUDT15 variants and NUDT15 nonvariant patients with IBD. Conversely, 6-TGNs were only associated with leukopenia in patients with IBD with NUDT15 nonvariants but not in patients with NUDT15 variants. This shows that DNA-TG does predict late leukopenia not only in NUDT15 nonvariants but also in NUDT15 variant patients highlighting the inclusive potential of DNA-TG measurements for patients with different ethnicities. Yang et al. [46] has also performed a study investigating early proactive monitoring of DNA-TG levels in patients with IBD. Late leukopenia was defined as a leukocyte count < 3.5 × 109/L over 2 months. In total, late leucopenia was observed in 15.6% (17/109) of NUDT15/TPMT wild-type and 64.1% (25/39) of intermediate metabolizers. DNA-TG levels were significantly higher in patients who developed leukopenia compared with those who did not develop leukopenia [394.9 ± 137.4 (median ± IQR, n = 42) versus 225.3 ± 139.6 (median ± IQR, n = 106) fmol/µg DNA; P = 4.9 × 10−13]. In TPMT/NUDT15 wild-type patients, the DNA-TG levels were also significantly higher in patients with leukopenia compared with those who did not [401.1 ± 222.9 (median ± IQR, n = 17) versus 221.1 ± 140.8 (median ± IQR, n = 92) fmol/µg DNA; P = 4.9 × 10−9). Furthermore, DNA-TG levels were also higher in variant patients who developed leukopenia compared with who did not develop leukopenia [382.5 ± 143.1 (median ± IQR, n = 25) versus 240.5 ± 248.3 (median ± IQR, n = 14) fmol/µg DNA; P = 0.024; 3 for TPMT*3C, 21 for NUDT15*3, 2 for NUDT15*6, and 14 for NUDT15*2]. With regards to 6-TGNs, in the entire cohort 6-TGN levels were significantly higher in patients who developed leukopenia compared with those who did not [322.4 ± 210.6 (median ± IQR, n = 42) versus 247.5 ± 182.5 (median ± IQR, n = 106) pmol/8 × 108 RBCs; P = 0.021). In wild-type patients, a similar result was found, in which 6-TGNs were higher in patients who developed leukopenia [333.7 ± 270.7 (median ± IQR, n = 17) versus 247.5 ± 162.9 (median ± IQR, n = 92) pmol/8 × 108 RBCs; P = 0.018]. However, similarly to the report of Zhu et al. [44], 6-TGNs were not able to predict leukopenia in patients with TPMT/NUDT15 variants [310.0 ± 180.6 (median ± IQR, n = 25) versus 249.9 ± 303.9 (median ± IQR, n = 14) pmol/8 × 108 RBCs; P = 0.55]. Again, this suggests that 6-TGNs are unable to predict leukopenia in variant-type patients, hindering the utility in especially patients with Asian and Hispanic ethnicities [16]. We have also performed a meta-analysis, which showed that DNA-TG is able to predict late leukopenia in patients with IBD. The DNA-TG levels were significantly higher in patients with leukopenia compared with those without leukopenia (mean difference 161.8, 95% CI 126.2–197.3, P < 0.00001).

3.3.4 Sinusoidal Obstruction Syndrome

Past discussions have highlighted concerns about the acute liver toxicity associated with thiopurine therapy, particularly for TG, especially in relation to conditions such as nodular regenerative hyperplasia and sinusoidal obstruction syndrome (SOS) [47,48,49]. Toksvang et al. [50] performed a subanalysis in the UKALL97 trial (11 cases with SOS) by measuring DNA-TG concentrations. In this trial, in which patients with ALL were randomized to either MP (n = 80) or TG (n = 52), no difference was found in DNA-TG levels between patients who received MP or TG. Furthermore, there was no difference in DNA-TG levels between patients who developed SOS compared with patients who did not develop SOS (HR 0.91 per 100 fmol/µg DNA-TG increase, 95% CI 0.76–1.09), using a Cox-regression analysis of relapse, death, second cancer, and SOS. This suggests that higher levels of DNA-TG are not directly related to SOS. They also investigated platelet counts as a derivative marker for hepatic angiopathy. In an analysis of the NOPHO ALL2008 study, no association was found between the median platelet count and median DNA-TG level during maintenance therapy [37]. Also, in the previously mentioned study by Larsen et al. [38], which added incremental low dosages of TG to MP and methotrexate (TEAM protocol) [39], no differences were found in platelet count compared with conventional protocol (NOPHO ALL2008). Hence, these studies suggest that higher levels DNA-TG are not directly related to SOS, which is consistent with the primary pathogenesis of SOS being due to a toxic effect of thioguanosine triphosphate on the sinusoidal endothelial cells rather than the circulating immune cells. [51]

3.3.5 Osteonecrosis

Osteonecrosis is a treatment-related toxicity in patients with ALL who are treated with MP/MTX maintenance therapy [52]. Toksvang et al. [53] prospectively registered osteonecrosis in 1234 patients with ALL who were treated with ALL2008 protocol (MP/MTX). After a median follow-up of 5.6 years, the cumulative incidence of osteonecrosis in their study was 2.7%. The median DNA-TG levels in the entire cohort were 384 fmol/μg DNA (IQR 211–367). Neither RBC 6-TGNs or DNA-TG were significantly associated with the risk of osteonecrosis in Cox models stratified by three age groups and adjusted for sex.

3.4 Relationship to Gene Variants of Thiopurine Metabolism

Numerous studies have examined the relationship between DNA-TG, thiopurine dosage, and RBC 6-TGN levels [23, 24, 37, 53,54,55], specifically focusing on gene variants involved in thiopurine metabolism such as TPMT, NUDT15, ITPA, NT5C2 and MRP4. This is noteworthy, as it may reveal DNA-TG’s potential to detect gene variants. Should this be the case, DNA-TG levels could potentially serve as a screening method to identify gene variants without the need for preliminary genetic testing. The significance of DNA-TG in relation to these genetic variants is detailed below.

3.4.1 TPMT

TPMT is an important enzyme in the metabolism pathways of thiopurines (Fig. 1). Raised levels of 6-TGNs are caused by a loss-of-function in the TPMT enzyme, which leads to an increased risk of developing leukopenia [56, 57]. The determination of TPMT genotype/phenotype can help guide physicians for dose adjustments of thiopurines, DNA-TG might be used to identify TPMT variant patients who did not have TPMT determination or may predict TPMT variant-related toxicities. Gerbek et al. [54] found significantly higher median DNA-TG levels in patients with ALL with TPMT low-activity variants compared with patients with wild-type TPMT (549 versus 364 fmol/µg, P = 0.001). Patients with TPMT-low variants also had higher RBC 6-TGN levels (491 versus 232 nmol/mmol Hb, P = 0.001), and DNA-TG levels positively correlated (rs = 0.37, P < 0.001) with RBC 6-TGN levels. Nielsen et al. [58] also reported that DNA-TG levels were higher for patients who were heterozygous for TPMT variants (760.9 fmol/µg DNA, IQR 568.7–890.3) compared with TPMT wild-type patients (492.7 fmol/µg DNA, IQR 382.1–634.6; P < 0.001). Patients carrying wild-type TPMT had higher TPMT activity compared with TPMT heterozygous genotypes [TPMT levels from maintenance therapy evaluated in patients who remained in remission at follow-up, N = 752, 16.6 (range 7.6−28.9) versus 9.6 (range 7.2−16.1) U/mL RBC; P < 0.001] but did not differ in their risk of leukemic relapse [58]. Ju et al. [42] measured DNA-TG levels in patients with ALL (n = 72) who were tested for TPMT and NUDT15 genotypes. No significant difference was found between TPMT variants (n = 3, *1/*3) and TPMT wild type in DNA-TG levels. No patients (n = 3) with a TPMT variant (*1/*3) developed leukopenia (WBC < 1500 WBC/μL). In their study, a significant positive correlation was found between MP intensity and DNA-TG concentration (P < 0.001). Furthermore, Zhu et al. [44] found that patients with IBD with TPMT*3C variants had higher levels of DNA-TG and RBC 6-TGNs compared with TPMT nonvariants, but no association was found between DNA-TG levels in TPMT*3C variants and leukopenia. However, in their study, the numbers of patients with TPMT variants was low.

3.4.2 ITPA

ITPA plays an important role in the hydrolysis of TITP to TIMP (Fig. 1), and the deficiency in ITPA leads to accumulation of TITP, which subsequently leads to toxicity [59]. A common single nucleotide polymorphism (SNP) in the ITPA gene is 94C > A, which was found to be associated with neutropenia in pediatric patients with ALL [60, 61]. Furthermore, from literature, patients with IBD with this ITPA variant had higher levels of 6-TGNs compared with patients with wild-type ITPA. In the same study, this ITPA genotype was associated with clinical effectiveness [62]. The effects of ITPA variants on DNA-TG levels were studied by Gerbek et al. [54] They reported significantly higher DNA-TG levels in patients with a low-activity ITPA variant compared with patients with wild-type ITPA [465 (IQR 419–558) versus 387 (IQR 287–448) fmol/µg DNA TG; P = 0.04). However, no differences were found between RBC 6-TGN levels (median 305 versus 260 nmol/mmol Hb, P = 0.086), RBC MethylMP levels (median 20,866 versus 15,902 nmol/mmol Hb, P = 0.12). In patients with low activity ITPA less TITP is reconverted to TIMP, resulting in higher levels of MeTITP. The authors suggested that this would lead to increased pools of methylated thiopurine metabolites, which might inhibit de novo purine synthesis and, thus, lead to lower levels of endogenous purines to compete with 6-TGN to be incorporated into DNA.

3.4.3 NUDT15

NUDT15 is a purine-specific nudix hydrolase that involves the hydrolysis of nucleosides-diphosphates [63]. NUDT15 is involved in the conversion of 6-TGTP and deoxy-TGTP metabolites to TGMP and deoxy-TGMP (Fig. 1) [64]. Deficiency in NUDT15 leads to relatively higher TGTP formation, which hypothetically promotes the formation of DNA-TG and inhibition of Rac1 [65]. Thus, NUDT15 may potentially increases the risk of developing leukopenia in patients with “normal” total 6-TGN levels, due to more abundant TGTP. Furthermore, a meta-analysis found that the risk of thiopurine-induced leukopenia was increased in 1136 NUDT15 R139C variant carriers compared with 4987 wild-type patients [odds ratio (OR) 6.9, 95% CI 5.2–9.1). [57] Thus, genotyping of NUDT15 seems clinically relevant, but the role of RBC 6-TGNs as a predictor for NUDT15-related leukopenia has not been clarified. However, it is anticipated that measurements of 6-TGN, which cannot detect NUDT15 variants, will also be unable to predict leukopenia. DNA-TG could present a compelling alternative to 6-TGN measurements since it targets the end-metabolite and is influenced by NUDT15 variants. Ju et al. [42] reported that the DNA-TG/MP dose ratio was significantly higher for patients carrying NUDT15 variants compared with patients with wild-type NUDT15 (P < 0.001). Four patients with NUDT15 variants (two *1/*2, one *1/*5, one *1/*6 genotype) developed leukopenia. The DNA-TG concentrations during these leukopenia episodes ranged from 27.8 to 504.8 fmol TG/μg DNA. Patients with NUDT15 variants had DNA-TG values of 235.9 (*1/*5), 68.7 (*1/*2), 62.0 (*1/*2), and 27.8 (*1/*6) (fmol TG/μg DNA). White blood cell count did not show a significant correlation with DNA-TG level (P = 0.093) or RBC 6-TGN (P = 0.425). Furthermore, Moriyama et al. [55] performed a study which compared RBC 6-TGN and DNA-TG levels in patients with ALL with NUDT15 variants. The mean RBC 6-TGN and DNA-TG levels were 134.1 pmol/4 × 108 RBC (range 0.46–315.5 pmol/4 × 108 RBC) and 442.8 fmol/μg DNA (range 78.1–1054.0 fmol/μg DNA). Patients with the low-activity NUDT15 diplotypes had the highest DNA-TG/dosage ratio, followed by the intermediate-activity diplotype group, and lowest in patients with wild-type NUDT15 (P = 4.0 × 10−9). A DNA-TG/6-TGN-ratio was used in their study and found that this ratio was significantly higher in NUDT15 deficient patients compared with NUDT15 wild-type patients (P = 3.6 × 10−9).

Choi et al. [23] validated an analytical assay that measured DNA-TG in Korean patients with ALL. They reported a moderate correlation between DNA-TG levels and RBC 6-TGNs in 257 measurements (ρ = 0.405, P < 0.0001), a moderate correlation between DNA-TG level and MP dose (ρ = 0.417, P < 0.0001). Furthermore, a higher DNA-TG/6-TGN-ratio was found in patients who were NUDT15 intermediate metabolizers (*1/*2, *1/*3) and indeterminate NUDT15 variants (*1/*5, *1/*6) compared with NUDT15 wild type. They also reported a linear relationship between DNA-TG and 6-TGN